from ai.backend import python @ pycontw2018
4 likes891 views
1. The document describes the role of an AI engineer at a computer vision startup called Umbo. It discusses developing and maintaining computer vision services, building machine learning pipelines, and lessons learned. 2. Key responsibilities of an AI engineer include developing and maintaining computer vision services, building machine learning pipelines to improve services and measure model performance, and debugging and refactoring code. 3. The author learned that domain knowledge is important, Python is a unifying language, and collaboration between researchers and engineers is necessary. The future may see more high-quality backend development to support machine learning services.
1 of 63
Downloaded 26 times
Ad
Recommended
Build and Host Real-world Machine Learning Services from Scratch @ pycontw2019
Build and Host Real-world Machine Learning Services from Scratch @ pycontw2019 Chun-Yu Tseng Joe Tseng, an AI engineer, discusses his experience building and hosting machine learning services, emphasizing the importance of data collection and model improvement in real-world applications. He shares insights on developing autonomous video security solutions, transition challenges from Python 2 to 3, and the adoption of modern technologies for enhanced infrastructure and performance. The talk highlights the need for a product mindset when deploying machine learning services to effectively meet real-world challenges.
Powering machine learning workflows with Apache Airflow and Python
Powering machine learning workflows with Apache Airflow and PythonTatiana Al-Chueyr This document provides an overview of using Apache Airflow to power machine learning workflows with Python. It discusses Airflow concepts like DAGs, operators, relationships and visualizations. It also covers installing Airflow, common issues experienced like debugging and versioning, and using Airflow for machine learning tasks like model building and hyperparameter tuning. Examples of Airflow pipelines for data ingestion and machine learning are demonstrated. The presenter's background and the BBC Datalab team are briefly introduced.
Scala in-practice-3-years by Patric Fornasier, Springr, presented at Pune Sca...
Scala in-practice-3-years by Patric Fornasier, Springr, presented at Pune Sca...Thoughtworks The document discusses the experience of a software engineering team at Springer after three years of using Scala for their content delivery platform. It highlights both positive aspects, such as increased productivity and the functional programming features of Scala, and negative aspects, including tool support issues and complexity. The future outlook suggests no immediate plans to move away from Scala, though Java 8 is considered an alternative as the company may explore polyglot programming with smaller teams and applications.
LINE NOW Scratch Card - From Nothing to Production in one month
LINE NOW Scratch Card - From Nothing to Production in one monthLINE Corporation The document summarizes the development of a scratch card game for LINE NOW by a team in one month. Key points:
- The team developed a scratch card reward game for an upcoming Moon Festival campaign in LINE NOW within one month using agile methods.
- They created an algorithm to randomly generate scratch cards that ensured prizes would be distributed evenly over time to avoid issues of running out of prizes or complaints from users.
- The architecture included LIFF and microservices with Redis to store card data, Kafka for events, and deployment to Kubernetes.
- Challenges included performance testing of the card generation and redemption APIs, scaling to handle campaign traffic peaks, and recovering when Kafka went down on the last
Hydrosphere.io for ODSC: Webinar on Kubeflow
Hydrosphere.io for ODSC: Webinar on KubeflowRustem Zakiev The document discusses a webinar focused on training and deploying machine learning models using tools like Kubeflow and Hydrosphere.io. It outlines a comprehensive machine learning workflow, detailing each step from research to model maintenance. Additionally, it explores automation processes and the use of containerization for efficient model management and deployment.
Runtime performance
Runtime performanceEliran Eliassy This document provides tips and tools for improving Angular application runtime performance. It discusses measuring performance using Benchmark.js, Chrome Dev Tools profiler, and @angular/benchpress. Specific optimizations covered include using the Ivy renderer, lazy loading modules, pre-rendering with Angular Universal, caching requests with service workers, immutable data structures, memoization, trackBy in *ngFor, virtual scrolling, RxJS operators, and controlling change detection. The document aims to help developers measure and improve the runtime performance of their Angular applications.
Large-Scale Training with GPUs at Facebook
Large-Scale Training with GPUs at FacebookFaisal Siddiqi This document discusses large-scale distributed training with GPUs at Facebook using their Caffe2 framework. It describes how Facebook was able to train the ResNet-50 model on the ImageNet dataset in just 1 hour using 32 GPUs with 8 GPUs each. It explains how synchronous SGD was implemented in Caffe2 using Gloo for efficient all-reduce operations. Linear scaling of the learning rate with increased batch size was found to work best when gradually warming up the learning rate over the first few epochs. Nearly linear speedup was achieved using this approach on commodity hardware.
From an idea to production: building a recommender for BBC Sounds
From an idea to production: building a recommender for BBC SoundsTatiana Al-Chueyr The document describes the process of developing and productionizing a recommendation engine for BBC Sounds. It discusses:
1) The initial challenge of replacing an outsourced recommendation engine and prototyping a new one using factorization machines. Qualitative user tests showed improved recommendations over the external provider.
2) Productionizing involved using Google Cloud Platform, Apache Airflow for workflows, Apache Beam for efficient data processing, and precomputing recommendations to serve 1500 requests/second with low latency.
3) Initial A/B tests found a 59% increase in interactions and 103% increase for under 35s using the new recommendation engine. Ongoing work includes optimizing costs and API performance.
Serverless in-action
Serverless in-actionAssaf Gannon The document outlines the implementation of a serverless architecture for an e-commerce shopping cart using GraphQL. It discusses the advantages of serverless systems, such as low costs and scalability, as well as specific tools and frameworks like Serverless.com and Apollo. The document also provides a step-by-step guide on setting up and deploying the solution on Google Cloud, along with best practices for production readiness.
Adding GraphQL to your existing architecture
Adding GraphQL to your existing architectureSashko Stubailo The document discusses adding GraphQL to existing architectures, emphasizing its incremental adoption and flexibility without necessitating a complete overhaul of existing systems. It highlights key considerations for implementing GraphQL, such as performance, client-side improvements, and optimizing backend load, along with practical guidance for getting started in production. The importance of community support and sharing experiences to enhance GraphQL implementation is also stressed.
Advanced Model Inferencing leveraging Kubeflow Serving, KNative and Istio
Advanced Model Inferencing leveraging Kubeflow Serving, KNative and IstioAnimesh Singh The document discusses advanced machine learning model serving using technologies like Knative, Istio, and Kubeflow, focusing on production deployment, monitoring, and maintenance of ML models. It highlights challenges in machine learning, introduces concepts such as distributed model training and hyperparameter optimization, and addresses the importance of explainability and monitoring in production environments. Additionally, it emphasizes collaboration across major contributors like IBM, Google, and Microsoft in evolving these technologies and practices.
Running Flink in Production: The good, The bad and The in Between - Lakshmi ...
Running Flink in Production: The good, The bad and The in Between - Lakshmi ...Flink Forward The document discusses the implementation and operation of Apache Flink at Lyft, highlighting various use cases such as dynamic pricing, fraud detection, and data pipeline management. It covers the lifecycle of a Flink job, challenges in production, integration with Kinesis streams, and improvements made for production stability and recoverability. Additionally, it mentions the release of the Flink operator for Kubernetes, which aims to streamline Flink job management and reduce downtime.
Cypher for Gremlin
Cypher for GremlinopenCypher This document discusses Cypher for Gremlin, which provides the ability to run Cypher queries against Gremlin databases. It covers the Gremlin and Cypher query languages, examples of translating Cypher to Gremlin, and the Java APIs for client-side and server-side translation of Cypher to Gremlin.
Meetup Angular.JS #12 Paris
Meetup Angular.JS #12 ParisSylvain Utard Sylvain Utard presents on Build Realtime Search. Realtime Search is a search API provider that processes 2 billion operations per month across 30+ servers. The backend is built in C++, analytics in Java, and the website in Rails. It supports 12 API clients in various languages. Realtime Search provides an average query time of less than 10ms across multiple datacenters globally. The REST API is CORS compliant for use in JavaScript applications and aims for an overall end-user latency of less than 100ms.
Microservices and serverless in python projects
Microservices and serverless in python projectsJose Manuel Ortega Candel The document discusses microservices and serverless architectures for Python projects. It covers microservices concepts and frameworks like asyncio and GraphQL. Serverless computing with AWS Lambda and frameworks like Zappa and Chalice are explained. Benefits of microservices include separation of concerns while serverless offers automatic scaling and lower costs. Frameworks make it easier to deploy Python code to AWS Lambda.
GraphQL across the stack: How everything fits together
GraphQL across the stack: How everything fits togetherSashko Stubailo The document discusses how GraphQL simplifies application development by providing a unified approach to client and server interactions, enabling developers to describe data requirements and capabilities effectively. It highlights the ecosystem surrounding Apollo and GraphQL, illustrating concepts such as caching, tracing, and schema stitching for improved performance and collaboration. The presentation emphasizes the importance of modularity and transparency in building complex applications with multiple frontends and backends.
Why I don’t want to develop iOS apps in Objective C
Why I don’t want to develop iOS apps in Objective CSeniorDevOnly The document discusses why the author prefers Swift over Objective-C for iOS app development, highlighting Swift's features such as optionals, immutability, generics, and a more elegant syntax. It also points out Swift's unique aspects like readonly properties and a rich ecosystem of libraries and frameworks like Reactive Cocoa and Vapor. The author argues that Swift offers modern programming benefits that enhance development efficiency and code quality.
Meteor MIT Tech Talk 9/18/14: Designing a New Platform For Modern Apps
Meteor MIT Tech Talk 9/18/14: Designing a New Platform For Modern AppsSashko Stubailo The document introduces Meteor, an open source JavaScript framework for building modern applications efficiently. Key features include instant response, reactive UI, fast load times, and cross-platform data access, enabled by technologies like DDP (Distributed Data Protocol) and Minimongo. It highlights the potential use cases for DDP and mentions job opportunities at Meteor.
Why UI Developers Love GraphQL - Sashko Stubailo, Apollo/Meteor
Why UI Developers Love GraphQL - Sashko Stubailo, Apollo/MeteorJon Wong The document discusses the advantages of using GraphQL for UI development, highlighting its efficiency in data fetching and the benefits of strong typing and built-in documentation. It compares GraphQL with REST APIs, emphasizing easier query management and caching strategies. Additionally, it explores tools and techniques for static analysis, autocomplete features, and the potential for client-side data handling and cache consistency.
Enabling Machine Learning with Apache Flink - Sherin Thomas, Lyft
Enabling Machine Learning with Apache Flink - Sherin Thomas, LyftFlink Forward The document discusses the integration of machine learning and streaming data processing using Apache Flink, detailing workflows for data preparation, model development, and deployment. It highlights techniques such as bootstrapping, feature engineering, and management of Flink on Kubernetes, emphasizing low latency operations and the importance of handling both historic and real-time data. Additionally, it notes the streamlined processes for feature creation and deployment, achieving operational efficiency and improved scaling capabilities.
GNAT Pro User Day: GNATdashboard - Tracking and Improving Software Quality
GNAT Pro User Day: GNATdashboard - Tracking and Improving Software QualityAdaCore The document discusses the gnatdashboard, focusing on tracking and improving software quality by managing technical debt from the beginning of projects. It outlines key factors affecting code quality, such as maintainability, complexity, and documentation, as well as tools like gnatmetric and sonarqube for monitoring these aspects. The gnatdashboard integrates these tools to provide a comprehensive view of code quality metrics and improvements over time.
The Apollo and GraphQL Stack
The Apollo and GraphQL StackSashko Stubailo The document discusses how Apollo and GraphQL can simplify application development amidst complexity from multiple frontends and backends. It highlights the benefits of using GraphQL, including fast development, better maintenance, and improved understanding of app data, supported by a suite of Apollo tools for both client and server sides. The presentation emphasizes the importance of collaboration, safety, and documenting workflows to enhance development experiences with GraphQL.
Coscup 2013 : Continuous Integration on top of hadoop
Coscup 2013 : Continuous Integration on top of hadoopWisely chen This document discusses implementing continuous integration (CI) for Hadoop projects. It describes problems with debugging and assessing performance of MapReduce jobs. The proposed solution is to set up a CI system for Hadoop that automates unit testing, performance testing, documentation generation and deployment. This allows developers to catch issues early before deploying to production and improves productivity. Demo examples are provided of the CI system failing and passing unit tests and assessing performance.
Raphael Amorim - Scrating React Fiber
Raphael Amorim - Scrating React FiberReact Conf Brasil The document discusses the evolution of React, highlighting the introduction of React Fiber, which enhances the rendering process through incremental rendering and asynchronous reconciliation. It addresses the impact of JavaScript engines on performance and how React can efficiently manage animation and rendering tasks. The author, Raphael Amorim, also shares his work in open source projects related to JavaScript and web technologies.
GraphQL over REST at Reactathon 2018
GraphQL over REST at Reactathon 2018Sashko Stubailo The document discusses the benefits of using GraphQL over REST for frontend development, highlighting how GraphQL simplifies API interactions and state management in React applications. It presents a strategy for implementing GraphQL on top of existing REST APIs, allowing developers to leverage the advantages without overhauling their current systems. Key tips include designing a fitting GraphQL schema based on frontend needs, integrating with existing REST endpoints, and optimizing for performance.
From zero to test in 60 seconds
From zero to test in 60 secondsHugh McCamphill This document discusses test automation and the importance of abstractions for strategic and maintainable automation. It recommends using page object modeling and selecting libraries like WebdriverIO, SuperTest, Allure, and Appium to integrate different testing needs. A sample test is shown using page objects to model the UI and compose tests in a browser-driven way while keeping the ability to swap steps to API testing. Overall abstractions through page objects and library integration are promoted for scalable and maintainable test automation.
SETCON'18 - Ilya labacheuski - GraphQL adventures
SETCON'18 - Ilya labacheuski - GraphQL adventuresNadzeya Pus The document describes a presentation by Ilya Labacheuski on implementing GraphQL using TypeScript, detailing its advantages over REST APIs, complexities, and server setup. It includes examples of schema definitions, queries, and potential migration strategies from REST to GraphQL, along with tooling recommendations for effective implementation. The presentation also discusses challenges faced during schema creation, such as handling complex data structures and ensuring appropriate type definitions.
ARI and AGI, a powerful combination
ARI and AGI, a powerful combinationJöran Vinzens This document discusses using Asterisk Gateway Interface (AGI), Asterisk Manager Interface (AMI), and Asterisk REST Interface (ARI) to build telephony features. It proposes using AGI for generic call setup/teardown and feature determination, and ARI apps running on an ARI proxy to control individual features during calls. This decouples features from Asterisk and isolates them, improving testability, replaceability, and avoiding direct developer interaction with Asterisk dialplan. The approach moves routing logic away from Asterisk and enables any feature via a low-level ARI API and high-level feature control.
Industrializing Machine learning pipelines
Industrializing Machine learning pipelinesGermain Tanguy This document discusses industrializing machine learning pipelines. It proposes a machine learning blueprint to improve collaboration between data scientists and engineers. The blueprint includes technical components like extracting, preprocessing, training, evaluating and predicting data. It defines roles for data scientists and engineers and recommends common tools and code to streamline workflows. The document also uses a case study of catalog categorization to illustrate how the blueprint was applied to classify videos using a machine learning pipeline with various technical components like TensorFlow, Apache Beam and BigQuery.
More Data, More Problems: Evolving big data machine learning pipelines with S...
More Data, More Problems: Evolving big data machine learning pipelines with S...Alex Sadovsky The document discusses the challenges and evolution of big data machine learning pipelines using Spark and Luigi, emphasizing the need for collaboration among various stakeholders in data science. It outlines the strengths and limitations of tools like Spark for data manipulation and modeling, alongside Luigi's capabilities for managing complex workflows. The presentation highlights the importance of automated machine learning services and efficient pipeline management to achieve successful data science outcomes.
More Related Content
What's hot (20)
Serverless in-action
Serverless in-actionAssaf Gannon The document outlines the implementation of a serverless architecture for an e-commerce shopping cart using GraphQL. It discusses the advantages of serverless systems, such as low costs and scalability, as well as specific tools and frameworks like Serverless.com and Apollo. The document also provides a step-by-step guide on setting up and deploying the solution on Google Cloud, along with best practices for production readiness.
Adding GraphQL to your existing architecture
Adding GraphQL to your existing architectureSashko Stubailo The document discusses adding GraphQL to existing architectures, emphasizing its incremental adoption and flexibility without necessitating a complete overhaul of existing systems. It highlights key considerations for implementing GraphQL, such as performance, client-side improvements, and optimizing backend load, along with practical guidance for getting started in production. The importance of community support and sharing experiences to enhance GraphQL implementation is also stressed.
Advanced Model Inferencing leveraging Kubeflow Serving, KNative and Istio
Advanced Model Inferencing leveraging Kubeflow Serving, KNative and IstioAnimesh Singh The document discusses advanced machine learning model serving using technologies like Knative, Istio, and Kubeflow, focusing on production deployment, monitoring, and maintenance of ML models. It highlights challenges in machine learning, introduces concepts such as distributed model training and hyperparameter optimization, and addresses the importance of explainability and monitoring in production environments. Additionally, it emphasizes collaboration across major contributors like IBM, Google, and Microsoft in evolving these technologies and practices.
Running Flink in Production: The good, The bad and The in Between - Lakshmi ...
Running Flink in Production: The good, The bad and The in Between - Lakshmi ...Flink Forward The document discusses the implementation and operation of Apache Flink at Lyft, highlighting various use cases such as dynamic pricing, fraud detection, and data pipeline management. It covers the lifecycle of a Flink job, challenges in production, integration with Kinesis streams, and improvements made for production stability and recoverability. Additionally, it mentions the release of the Flink operator for Kubernetes, which aims to streamline Flink job management and reduce downtime.
Cypher for Gremlin
Cypher for GremlinopenCypher This document discusses Cypher for Gremlin, which provides the ability to run Cypher queries against Gremlin databases. It covers the Gremlin and Cypher query languages, examples of translating Cypher to Gremlin, and the Java APIs for client-side and server-side translation of Cypher to Gremlin.
Meetup Angular.JS #12 Paris
Meetup Angular.JS #12 ParisSylvain Utard Sylvain Utard presents on Build Realtime Search. Realtime Search is a search API provider that processes 2 billion operations per month across 30+ servers. The backend is built in C++, analytics in Java, and the website in Rails. It supports 12 API clients in various languages. Realtime Search provides an average query time of less than 10ms across multiple datacenters globally. The REST API is CORS compliant for use in JavaScript applications and aims for an overall end-user latency of less than 100ms.
Microservices and serverless in python projects
Microservices and serverless in python projectsJose Manuel Ortega Candel The document discusses microservices and serverless architectures for Python projects. It covers microservices concepts and frameworks like asyncio and GraphQL. Serverless computing with AWS Lambda and frameworks like Zappa and Chalice are explained. Benefits of microservices include separation of concerns while serverless offers automatic scaling and lower costs. Frameworks make it easier to deploy Python code to AWS Lambda.
GraphQL across the stack: How everything fits together
GraphQL across the stack: How everything fits togetherSashko Stubailo The document discusses how GraphQL simplifies application development by providing a unified approach to client and server interactions, enabling developers to describe data requirements and capabilities effectively. It highlights the ecosystem surrounding Apollo and GraphQL, illustrating concepts such as caching, tracing, and schema stitching for improved performance and collaboration. The presentation emphasizes the importance of modularity and transparency in building complex applications with multiple frontends and backends.
Why I don’t want to develop iOS apps in Objective C
Why I don’t want to develop iOS apps in Objective CSeniorDevOnly The document discusses why the author prefers Swift over Objective-C for iOS app development, highlighting Swift's features such as optionals, immutability, generics, and a more elegant syntax. It also points out Swift's unique aspects like readonly properties and a rich ecosystem of libraries and frameworks like Reactive Cocoa and Vapor. The author argues that Swift offers modern programming benefits that enhance development efficiency and code quality.
Meteor MIT Tech Talk 9/18/14: Designing a New Platform For Modern Apps
Meteor MIT Tech Talk 9/18/14: Designing a New Platform For Modern AppsSashko Stubailo The document introduces Meteor, an open source JavaScript framework for building modern applications efficiently. Key features include instant response, reactive UI, fast load times, and cross-platform data access, enabled by technologies like DDP (Distributed Data Protocol) and Minimongo. It highlights the potential use cases for DDP and mentions job opportunities at Meteor.
Why UI Developers Love GraphQL - Sashko Stubailo, Apollo/Meteor
Why UI Developers Love GraphQL - Sashko Stubailo, Apollo/MeteorJon Wong The document discusses the advantages of using GraphQL for UI development, highlighting its efficiency in data fetching and the benefits of strong typing and built-in documentation. It compares GraphQL with REST APIs, emphasizing easier query management and caching strategies. Additionally, it explores tools and techniques for static analysis, autocomplete features, and the potential for client-side data handling and cache consistency.
Enabling Machine Learning with Apache Flink - Sherin Thomas, Lyft
Enabling Machine Learning with Apache Flink - Sherin Thomas, LyftFlink Forward The document discusses the integration of machine learning and streaming data processing using Apache Flink, detailing workflows for data preparation, model development, and deployment. It highlights techniques such as bootstrapping, feature engineering, and management of Flink on Kubernetes, emphasizing low latency operations and the importance of handling both historic and real-time data. Additionally, it notes the streamlined processes for feature creation and deployment, achieving operational efficiency and improved scaling capabilities.
GNAT Pro User Day: GNATdashboard - Tracking and Improving Software Quality
GNAT Pro User Day: GNATdashboard - Tracking and Improving Software QualityAdaCore The document discusses the gnatdashboard, focusing on tracking and improving software quality by managing technical debt from the beginning of projects. It outlines key factors affecting code quality, such as maintainability, complexity, and documentation, as well as tools like gnatmetric and sonarqube for monitoring these aspects. The gnatdashboard integrates these tools to provide a comprehensive view of code quality metrics and improvements over time.
The Apollo and GraphQL Stack
The Apollo and GraphQL StackSashko Stubailo The document discusses how Apollo and GraphQL can simplify application development amidst complexity from multiple frontends and backends. It highlights the benefits of using GraphQL, including fast development, better maintenance, and improved understanding of app data, supported by a suite of Apollo tools for both client and server sides. The presentation emphasizes the importance of collaboration, safety, and documenting workflows to enhance development experiences with GraphQL.
Coscup 2013 : Continuous Integration on top of hadoop
Coscup 2013 : Continuous Integration on top of hadoopWisely chen This document discusses implementing continuous integration (CI) for Hadoop projects. It describes problems with debugging and assessing performance of MapReduce jobs. The proposed solution is to set up a CI system for Hadoop that automates unit testing, performance testing, documentation generation and deployment. This allows developers to catch issues early before deploying to production and improves productivity. Demo examples are provided of the CI system failing and passing unit tests and assessing performance.
Raphael Amorim - Scrating React Fiber
Raphael Amorim - Scrating React FiberReact Conf Brasil The document discusses the evolution of React, highlighting the introduction of React Fiber, which enhances the rendering process through incremental rendering and asynchronous reconciliation. It addresses the impact of JavaScript engines on performance and how React can efficiently manage animation and rendering tasks. The author, Raphael Amorim, also shares his work in open source projects related to JavaScript and web technologies.
GraphQL over REST at Reactathon 2018
GraphQL over REST at Reactathon 2018Sashko Stubailo The document discusses the benefits of using GraphQL over REST for frontend development, highlighting how GraphQL simplifies API interactions and state management in React applications. It presents a strategy for implementing GraphQL on top of existing REST APIs, allowing developers to leverage the advantages without overhauling their current systems. Key tips include designing a fitting GraphQL schema based on frontend needs, integrating with existing REST endpoints, and optimizing for performance.
From zero to test in 60 seconds
From zero to test in 60 secondsHugh McCamphill This document discusses test automation and the importance of abstractions for strategic and maintainable automation. It recommends using page object modeling and selecting libraries like WebdriverIO, SuperTest, Allure, and Appium to integrate different testing needs. A sample test is shown using page objects to model the UI and compose tests in a browser-driven way while keeping the ability to swap steps to API testing. Overall abstractions through page objects and library integration are promoted for scalable and maintainable test automation.
SETCON'18 - Ilya labacheuski - GraphQL adventures
SETCON'18 - Ilya labacheuski - GraphQL adventuresNadzeya Pus The document describes a presentation by Ilya Labacheuski on implementing GraphQL using TypeScript, detailing its advantages over REST APIs, complexities, and server setup. It includes examples of schema definitions, queries, and potential migration strategies from REST to GraphQL, along with tooling recommendations for effective implementation. The presentation also discusses challenges faced during schema creation, such as handling complex data structures and ensuring appropriate type definitions.
ARI and AGI, a powerful combination
ARI and AGI, a powerful combinationJöran Vinzens This document discusses using Asterisk Gateway Interface (AGI), Asterisk Manager Interface (AMI), and Asterisk REST Interface (ARI) to build telephony features. It proposes using AGI for generic call setup/teardown and feature determination, and ARI apps running on an ARI proxy to control individual features during calls. This decouples features from Asterisk and isolates them, improving testability, replaceability, and avoiding direct developer interaction with Asterisk dialplan. The approach moves routing logic away from Asterisk and enables any feature via a low-level ARI API and high-level feature control.
Similar to from ai.backend import python @ pycontw2018 (20)
Industrializing Machine learning pipelines
Industrializing Machine learning pipelinesGermain Tanguy This document discusses industrializing machine learning pipelines. It proposes a machine learning blueprint to improve collaboration between data scientists and engineers. The blueprint includes technical components like extracting, preprocessing, training, evaluating and predicting data. It defines roles for data scientists and engineers and recommends common tools and code to streamline workflows. The document also uses a case study of catalog categorization to illustrate how the blueprint was applied to classify videos using a machine learning pipeline with various technical components like TensorFlow, Apache Beam and BigQuery.
More Data, More Problems: Evolving big data machine learning pipelines with S...
More Data, More Problems: Evolving big data machine learning pipelines with S...Alex Sadovsky The document discusses the challenges and evolution of big data machine learning pipelines using Spark and Luigi, emphasizing the need for collaboration among various stakeholders in data science. It outlines the strengths and limitations of tools like Spark for data manipulation and modeling, alongside Luigi's capabilities for managing complex workflows. The presentation highlights the importance of automated machine learning services and efficient pipeline management to achieve successful data science outcomes.
Building a high-performance, scalable ML & NLP platform with Python, Sheer El...
Building a high-performance, scalable ML & NLP platform with Python, Sheer El...Pôle Systematic Paris-Region This document summarizes the development of Lore's machine learning and NLP platform using Python. It started as a monolithic Python server but evolved into a microservices architecture using Docker, Kubernetes, and Celery for parallelization. Key lessons included using DevOps tools like Docker for development and deployment, Celery to parallelize tasks, and wrapping services to improve modularity, flexibility, and performance. The platform now supports multiple products and consulting work in a scalable and maintainable way.
AI and Innovations on AWS
AI and Innovations on AWSAdrian Hornsby The document discusses artificial intelligence services available on AWS, including text-to-speech (Polly), image analysis (Rekognition), speech and natural language processing (Lex), machine learning (ML), deep learning frameworks, elastic GPUs on EC2, and FPGAs. It provides examples of how customers use these services for applications like facial recognition, object detection, predictive analytics, and autonomous systems. The document promotes AWS as providing the tools, compute, and services to scale AI applications.
A survey on Machine Learning In Production (July 2018)
A survey on Machine Learning In Production (July 2018)Arnab Biswas The document discusses the complexities and workflows associated with deploying machine learning models in production, emphasizing the differences between model and software development. It covers processes such as data generation, model training, evaluation, deployment, and monitoring, along with tools like Amazon SageMaker and Uber's Michelangelo for managing ML workflows. Best practices for ensuring model quality and robustness, including versioning, monitoring, and continuous training, are also highlighted.
Machine learning and big data @ uber a tale of two systems
Machine learning and big data @ uber a tale of two systemsZhenxiao Luo This document discusses machine learning and big data at Uber. It describes Uber's mission to provide reliable transportation everywhere. It then discusses Uber's use of Kafka for data infrastructure and Hadoop/Spark for batch processing and machine learning. The document outlines pain points with the current system and proposes addressing them by building machine learning capabilities directly into Presto, an interactive SQL query engine, to improve efficiency and allow machine learning and data analytics to leverage shared resources. Key benefits of this approach are reducing complexity for data scientists and improving performance of machine learning pipelines.
Image Classification and Retrieval on Spark
Image Classification and Retrieval on SparkGianvito Siciliano This document discusses using Spark Mbuto to build machine learning pipelines for image classification and retrieval. It provides an overview of classification and retrieval problems and the logic behind building pipelines for each. Key steps in the pipelines include feature extraction, building a codebook/dictionary, training and evaluating classifiers. Classification models discussed include KNN and neural networks. The document outlines building image pipelines in Spark Mbuto through sequential Spark jobs that share a context.
Devoxx: Building AI-powered applications on AWS
Devoxx: Building AI-powered applications on AWSAdrian Hornsby The document discusses building AI-powered applications on AWS, covering topics such as the use of AWS Lambda for event-driven architecture and serverless computing. It highlights various AWS AI services like Amazon Polly for text-to-speech and Amazon Rekognition for image analysis, emphasizing integration potential for developers. The session aims to provide historical context, practical applications, and showcase AI capabilities available through AWS.
03_aiops-1.pptx
03_aiops-1.pptxFarazulHoda2 This document discusses several tools that can be used to help manage and track machine learning projects from prototyping to production deployment. It describes Neptune, MLFlow, Kubeflow, and Pachyderm. Neptune and MLFlow are experiment trackers that can log parameters, results and artifacts from machine learning runs. Kubeflow provides notebooks, pipelines and model deployment on Kubernetes. Pachyderm is a versioning system for data pipelines that provides data provenance and tracks changes through containers on Kubernetes.
Canada DevOps Summit 2020 Presentation Nov_03_2020
Canada DevOps Summit 2020 Presentation Nov_03_2020Varun Manik Varun Kumar is a senior consultant at Deloitte with over 8 years of experience in cloud engineering and DevOps, holding several AWS certifications and a master's degree in computer science. He specializes in machine learning, cloud migrations, and automation, having recently led a cloud migration project for a media company in Southeast Asia. Varun is also an active technical blogger and trainer, contributing to the community through various platforms and training sessions.
AI meets Big Data
AI meets Big DataJan Wiegelmann This document discusses artificial intelligence and machine learning. It provides a brief history of AI from the Perceptron model in 1958 to modern deep learning approaches. It then discusses several applications of machine learning like image classification, medical diagnosis, and autonomous vehicles. It also discusses challenges like distributed machine learning and hidden technical debt. Finally, it provides examples of how AI can be applied to commerce and automotive use cases.
Azure Engineering MLOps
Azure Engineering MLOps Minesh A. Jethva This document provides an overview of Engineering MLOps, which refers to rapidly building, testing, and managing production-ready machine learning life cycles at scale. It discusses the evolution of software development and infrastructure that led to the need for MLOps. It also summarizes the key concepts and workflow of MLOps, including discussing a use case to illustrate the MLOps process.
AWS Machine Learning & Google Cloud Machine Learning
AWS Machine Learning & Google Cloud Machine LearningSC5.io This document provides an overview and comparison of machine learning services from AWS and Google Cloud. It begins with introductions of the speaker and agenda. It then provides background on machine learning and the three main types (supervised, unsupervised, reinforcement learning). It discusses how cloud services can provide on-demand compute for machine learning. It gives a breakdown of specific machine learning services from Google Cloud (such as Cloud ML Engine, Vision, Translation) and AWS (such as Machine Learning, Lex, Rekognition). It provides an example of pricing cloud infrastructure. Finally, it demonstrates building a multi-class classifier on the Iris dataset using logistic regression with both Google Cloud ML Engine and AWS Machine Learning.
My Resume
My ResumeDeepak Kumar M. Deepak Kumar has experience as a Machine Learning Engineer and Software Engineer Intern. He has worked at companies including Spline.ai, Betterhalf.ai, and TWB Inc., where he built and deployed machine learning models for tasks like disease prediction, personality analysis, and ad targeting. Deepak has skills in Python, JavaScript, machine learning frameworks like TensorFlow and Keras, and web technologies like Node.js, Flask, and Bootstrap. He holds a B.Tech. in Electrical and Electronics Engineering from Christ University.
Hadoop Infrastructure @Uber Past, Present and Future
Hadoop Infrastructure @Uber Past, Present and FutureDataWorks Summit The document presents an overview of Uber's data infrastructure evolution from 2014 to present, highlighting challenges such as scalability and reliability in data management. It discusses the integration of various tools like Hadoop, Mesos, and a unified scheduler for efficient resource management and data processing. Future directions include improvements in batch workload support and the development of the Peloton framework for enhanced scheduling capabilities.
Certification Study Group - NLP & Recommendation Systems on GCP Session 5
Certification Study Group - NLP & Recommendation Systems on GCP Session 5gdgsurrey Session 5, led by Chengcheng Tan, focuses on professional machine learning engineering, including an overview of various machine learning concepts and tools. The session covers topics such as the evolution of natural language processing, the architecture of large language models (LLMs), and practical applications of the Gemini API. It also explores ML pipeline automation, monitoring, optimization, and introduces TensorFlow Extended (TFX) for managing machine learning workflows.
Python in Industry
Python in IndustryDharmit Shah The document outlines career opportunities for Python programmers, including its use in various industries such as web development, data analysis, and machine learning. It discusses Python's history, popular frameworks like Django and Flask, and emphasizes the growing demand for skills in data analysis and automation. Additionally, it provides resources for improving Python skills, including online courses and community involvement.
AWS DevDay Seoul 2017 - Keynote
AWS DevDay Seoul 2017 - Keynote Amazon Web Services Korea The document discusses future trends in technology including artificial intelligence, machine learning, and cloud computing. It notes that AI/ML will drive new interaction channels and integration, connectivity, and openness will be key. Everything will increasingly be connected and able to be accessed through APIs. Deep learning will enable more human-like AI capabilities. It promotes experimenting with new technologies.
Real-Time AI: Designing for Low Latency and High Throughput - Dr. Sergei Izra...
Real-Time AI: Designing for Low Latency and High Throughput - Dr. Sergei Izra...Sri Ambati The document discusses the design and implementation of real-time machine learning applications, particularly at Beeswax, emphasizing the importance of low latency and high throughput. It outlines definitions of 'real time,' the trade-offs between batch and real-time scoring, and various machine learning model deployment options. Key considerations include system complexity, data handling, and ensuring optimal performance in producing predictions within milliseconds.
MACHINE LEARNING AUTOMATIONS PIPELINE WITH CI/CD
MACHINE LEARNING AUTOMATIONS PIPELINE WITH CI/CDIRJET Journal This document discusses integrating machine learning pipelines with continuous integration and continuous deployment (CI/CD) tools to automate machine learning workflows. It proposes using DevOps tools like Jenkins, Docker, and GitHub to build a CI/CD pipeline for machine learning. The pipeline would include steps for data preprocessing, model training, evaluation, and deployment. Continuous integration would involve regular code updates and testing. Continuous deployment would push trained models to production for monitoring. The goal is to reduce costs and resources needed for machine learning projects through automation with DevOps practices like CI/CD.
Building a high-performance, scalable ML & NLP platform with Python, Sheer El...
Building a high-performance, scalable ML & NLP platform with Python, Sheer El...Pôle Systematic Paris-Region
Ad
More from Chun-Yu Tseng (19)
驚呆了！這些 Python 題目刷掉 7 成面試者
驚呆了！這些 Python 題目刷掉 7 成面試者Chun-Yu Tseng 此文档讨论了一位面试官在技术面试中如何通过快问快答的方式快速评估面试者的技术水平。面试过程中的题目设计旨在筛选出大约70%的不合适候选人，同时强调面试者还需准备其他与岗位相关的知识。结论是，简单的问题能够有效过滤人选，且Python作为热门语言并未导致对编程语言的深入探讨。
5 minutes 介紹台南拍 (2022 ver.)
5 minutes 介紹台南拍 (2022 ver.)Chun-Yu Tseng 文档回顾了 tainan.py 的历史和发展，包括活动由 2013 年至 2020 年间的组织情况。主办方分享了他们的经历和未来的期望，强调继续与其他技术社区合作并鼓励投稿。文中感谢了多个支持的社群和场地，对于未来活动的持续举办充满期待。
Tenet: 2020 Taiwan PyCon Host We How
Tenet: 2020 Taiwan PyCon Host We HowChun-Yu Tseng The document outlines the planning process and events leading up to the 2020 Taiwan PyCon, detailing meetings, ticket sales, sponsorships, and preparations. Key milestones include achieving ticket sales goals and organizing hybrid presentations. The author invites others to join the preparation team for the following year's event.
導讀 Clean Code
導讀 Clean CodeChun-Yu Tseng Le document traite des principes de codage propre et efficace, mettant l'accent sur des pratiques telles que les design patterns et le développement dirigé par les tests (TDD). Il mentionne également des outils et des ressources pour aider à maintenir un code de haute qualité, comme flake8 et la norme PEP8. Les concepts de programmation agile et les différentes approches de développement logiciel sont également abordés.
Essential TDD @ pycontw2016
Essential TDD @ pycontw2016Chun-Yu Tseng This document provides an overview of a presentation on essential test-driven development (TDD) for Python developers. The presentation covers TDD principles and techniques, including writing tests before code, running tests frequently during development, and using tests to drive code design. It also discusses acceptance test-driven development (ATDD) and behavior-driven development (BDD), which apply TDD principles at a broader scope. The presentation concludes with recommendations for adopting a TDD mindset and choosing appropriate development tools and environments to support the TDD process.
從蟒蛇到神龍 - 從 1 接關繼續打造爬蟲程式
從蟒蛇到神龍 - 從 1 接關繼續打造爬蟲程式Chun-Yu Tseng Le document fournit un guide technique sur l'utilisation de divers modules Python pour le crawling et le scraping web, notamment des outils comme 'whois', 'requests', et 'BeautifulSoup'. Il détaille les étapes à suivre, y compris l'analyse des fichiers robots.txt, l'utilisation de regex pour le traitement de modèles, et les méthodes pour contourner des obstacles comme les CAPTCHA. Enfin, il mentionne des solutions pour le rendu JavaScript et l'automatisation du navigateur.
Agile x API x Documentation @ NGO [[MOPCON2015]]
Agile x API x Documentation @ NGO [[MOPCON2015]]Chun-Yu Tseng Le document décrit un environnement de développement agile utilisant diverses technologies comme Python, JavaScript, et des frameworks pour la création de projets open source. Il mentionne le processus Scrum incluant des réunions planifiées, des revues de code et des tests, ainsi que des outils de déploiement et de CI pour assurer la qualité du code. Des détails spécifiques sur les pratiques de test et la documentation API sont également fournis.
快快樂樂成為 Coding Ninja (by pytest) @ PyConAPAC2015
快快樂樂成為 Coding Ninja (by pytest) @ PyConAPAC2015Chun-Yu Tseng 本文探讨了如何使用pytest进行软件测试，提高开发者的测试思维。它介绍了pytest的基本功能与优点，以及为什么测试对开发者至关重要。文中还提到在写测试过程中可能遇到的问题及其解决方案。
程式 x 設計 @ MOPCON 2014
程式 x 設計 @ MOPCON 2014Chun-Yu Tseng 本文件探讨了程式设计与设计思维之间的关系，强调每个开发者都应具备设计思维。通过分析设计原则和心理学，内容提供了关于如何在编程中实现可用性和用户体验的指导。结尾总结指出，设计不仅是视觉上的，而是贯穿于各个开发阶段的重要思考。
介紹 MOSUT @ 2014.09.26 自由軟體開發與社群發展
介紹 MOSUT @ 2014.09.26 自由軟體開發與社群發展Chun-Yu Tseng 本文介绍了成大資工的社群活動mosut，强调自学和技能提升的重要性以及参与社群的好处。作者分享了参加mosut的理由、活动形式和内容，并指出了成大資工在课程和教学方面的不足。最后，强调了通过社群学习和交流可以突破传统教育的限制。
OO x Python @ Tainan.py x MOSUT x FP 2014.09.27
OO x Python @ Tainan.py x MOSUT x FP 2014.09.27Chun-Yu Tseng 这份文档讨论了面向对象编程（OOP）的起源、历史和多种语言（如C++、Python、Java、Ruby）的演变，强调不同观点对OOP的理解。通过问答形式，引导听众思考OOP的本质及其教学资源，包括在线课程和参考文献。最后提到Python的多样性和其在数据科学领域的受欢迎程度。
用路人小幫手 x 回報大改造 @ 2014 台南黑客松
用路人小幫手 x 回報大改造 @ 2014 台南黑客松Chun-Yu Tseng 本文件讨论了用路人小助手与道路回报系统的改进，提到当前的施工信息与回报系统存在的种种问题，包括信息分散与界面不友好。建议通过开放数据 API 进行系统整合，提升用户体验并加速问题的反馈处理。最终目标是实现即时有效的道路维护与回报过程。
Testing in Python @ Kaosiung.py 2014.05.26
Testing in Python @ Kaosiung.py 2014.05.26Chun-Yu Tseng 文档介绍了 Python 中的各种测试框架和工具，以及测试驱动开发 (TDD) 的实践。涵盖了 unittest、nose、py.test 等框架，并讨论了 TDD 的模式和成本。此外，还提到了即将举行的 Tainan.py 会议，鼓励社区成员参与。
Tip for Editors
Tip for EditorsChun-Yu Tseng 本文讨论了Python中的测试方法，包括单元测试、集成测试和功能测试。强调了有效测试的重要性，并介绍了使用断言和其他工具来优化测试流程。最后指出了不同测试类型的特性和适用场景。
PyConAPAC2014 BoF Introduction
PyConAPAC2014 BoF IntroductionChun-Yu Tseng The document discusses the status of 'Birds of a Feather' (BoF) sessions at PyCon APAC 2014, asserting that they are not dead and emphasizing the importance of sharing food, specifically pizza, during these gatherings. It highlights issues encountered with room accessibility and testing logistics, illustrating the challenges of hosting productive sessions. The conclusion reiterates enthusiasm for BoF sessions and the communal aspect of eating together.
PyConAPAC2014 Tainan.py 介紹
PyConAPAC2014 Tainan.py 介紹Chun-Yu Tseng 文档介绍了一项在台南举行的 Python 社群聚会活动，该活动自 2013 年以来定期举办，包含演讲、讨论和志愿者招募。聚会内容包括多种编程主题和实用技能的分享，并鼓励参与者投稿和互动。下次聚会定于 5 月 31 日，主题为 Python 测试和 TDD。
暖場 @ Tainan.py 2013.11.30
暖場 @ Tainan.py 2013.11.30 Chun-Yu Tseng 本文件欢迎参与聚会，提到缴交保護費、聚会场地提供小吃和录影支援等安排。介绍了聚会的内容，包括演讲和工作坊，还呼吁志工参与和投稿。文件还简要提到了聚会的背景与历史，展示社区活动的活跃性。
API 文件大亂鬥 @ MOPCON 2013
API 文件大亂鬥 @ MOPCON 2013Chun-Yu Tseng 文档探讨了撰写API文件的多种方法及何时可能不需要撰写文件，包括使用工具、自动生成以及换取同事自主查看代码等方式。此外，作者强调了维护文件的复杂性和必要性，并介绍了Python框架的自动文档生成。最后，鼓励读者分享其他方法，并提供聚会信息。
暖場與 Web 相關的小玩具 @ Tainan.py 2013.09.28
暖場與 Web 相關的小玩具 @ Tainan.py 2013.09.28Chun-Yu Tseng 该文档介绍了九月份在 Tainan.py 举办的聚会和相关活动，包含与 mosut 社群的合作。会议内容涵盖了 Python 使用的多个小工具及其应用示例，并安排了众多演讲者分享经验和技术。聚会主要在成大和南大举行，并设有社交环节。
Ad
Recently uploaded (20)
Download Adobe Illustrator Crack free for Windows 2025?
Download Adobe Illustrator Crack free for Windows 2025?grete1122g COPY & PASTE LINK 👉👉👉 https://click4pc.com/after-verification-click-go-to-download-page/
Adobe Illustrator CC Crack will make your illustration level upgrade by moving toward real modification purpose using adobe pro tools there ..
Folding Cheat Sheet # 9 - List Unfolding 𝑢𝑛𝑓𝑜𝑙𝑑 as the Computational Dual of ...
Folding Cheat Sheet # 9 - List Unfolding 𝑢𝑛𝑓𝑜𝑙𝑑 as the Computational Dual of ...Philip Schwarz For newer version see https://fpilluminated.org/deck/264
In this deck we look at the following:
* How unfolding lists is the computational dual of folding lists
* Different variants of the function for unfolding lists
* How they relate to the iterate function
Automated Testing and Safety Analysis of Deep Neural Networks
Automated Testing and Safety Analysis of Deep Neural NetworksLionel Briand Keynote address at Internetware 2025, co-located with FSE 2025
Top Time Tracking Solutions for Accountants
Top Time Tracking Solutions for Accountantsoliviareed320 This presentation explores the importance of time tracking for accountants and introduces modern solutions that help streamline billable hours, improve productivity, and eliminate manual tracking errors. Learn how time tracking software tailored for accounting professionals can simplify client billing, ensure compliance, and enhance profitability.
From key challenges in time management to top features to look for in a solution, this presentation offers valuable insights and practical tips for solo accountants, finance teams, and accounting firms looking to optimize their time-tracking workflows.https://www.invoicera.com/blog/invoicing/best-time-tracking-software-for-accountants/
ERP Systems in the UAE: Driving Business Transformation with Smart Solutions
ERP Systems in the UAE: Driving Business Transformation with Smart Solutionsdheeodoo ERP solutions are driving innovative change for businesses across the UAE through improved efficiency, reduced operational costs, and provide real-time visibility of operational data. By offering native Arabic language support, VAT compliance, and some level of cloud readiness, ERP solutions also supports and are consistent with the initiative for Smart Dubai 2030. Odoo is an example of a flexible & scalable solution. If you're looking for expert support with implementation contact Banibro. www.banibro.com
Why Edge Computing Matters in Mobile Application Tech.pdf
Why Edge Computing Matters in Mobile Application Tech.pdfIMG Global Infotech This PDF explores the critical role of edge computing in modern mobile application technology. Learn how edge enhances app speed, reduces latency, improves data privacy, and supports real-time performance. Ideal for developers, tech strategists, and mobile innovators looking to stay ahead in the evolving mobile ecosystem.
Why Every Growing Business Needs a Staff Augmentation Company IN USA.pdf
Why Every Growing Business Needs a Staff Augmentation Company IN USA.pdfmary rojas U.S. staff augmentation companies excel at providing IT professionals with expertise in development, cybersecurity, AI, and more. Whether you need to fill a short-term gap or enhance your team’s capabilities long-term, these firms deliver pre-vetted talent tailored to your tech stack and industry. With an understanding of both global technologies and local compliance, U.S.-based IT staff augmentation bridges the gap between talent shortage and project demand efficiently.
Y - Recursion The Hard Way GopherCon EU 2025
Y - Recursion The Hard Way GopherCon EU 2025Eleanor McHugh In most modern programming languages, including Go, it's possible for a function to call itself recursively. It's such a mainstream technique that we've come to take it for granted.
But what if functions couldn't call themselves?
There's a branch of mathematics called Combinatorics which answers just these kinds of questions, and this presentation takes these ideas and phrases them in Go to teach a deeper view of computation through the medium of code.
Zoho Creator Solution for EI by Elsner Technologies.docx
Zoho Creator Solution for EI by Elsner Technologies.docxElsner Technologies Pvt. Ltd. Working with Elsner’s Zoho Creator experts, Educational Initiatives implemented a custom asset management system that: Our Zoho Creator solutions have made real changes across the company, and each team has felt the difference.
How Automation in Claims Handling Streamlined Operations
How Automation in Claims Handling Streamlined OperationsInsurance Tech Services Claims processing automation reduces manual work, cuts errors, and speeds up resolutions. Using AI, RPA, and OCR, it enables faster data handling and smoother workflows—boosting efficiency and customer satisfaction. Visit https://www.damcogroup.com/insurance/claims-processing-automation for more details!
Introduction to Agile Frameworks for Product Managers.pdf
Introduction to Agile Frameworks for Product Managers.pdfAli Vahed As a Product Manager, having a solid understanding of Agile frameworks is essential, whether you are joining an established team or helping shape a new one.
Agile is not just a delivery method; it is a mind set that directly impacts how products are built, iterated, and brought to market. Knowing the strengths and trade-offs of each framework enables you to collaborate more effectively with teams, foster alignment, and drive meaningful outcomes.
Simply put, Agile fluency helps you lead with clarity in fast-paced, ever-changing environments.
The next few slides are simple ‘STUDY CARDS’ to help you learn or refresh your understanding of a few popular Agile frameworks.
From Data Preparation to Inference: How Alluxio Speeds Up AI
From Data Preparation to Inference: How Alluxio Speeds Up AIAlluxio, Inc. Alluxio Webinar
June 17, 2025
For more Alluxio Events: https://www.alluxio.io/events/
Speaker:
Jingwen Ouyang (Sr. Product Manager @ Alluxio)
In this talk, Jingwen Ouyang, Senior Product Manager at Alluxio, will share how Alluxio make it easy to share and manage data from any storage to any compute engine in any environment with high performance and low cost for your model training, model inference, and model distribution workload.
Humans vs AI Call Agents - Qcall.ai's Special Report
Humans vs AI Call Agents - Qcall.ai's Special ReportUdit Goenka Humans vs Agentic AI Cost Comparison: Cut 85% Support Costs
TL;DR: Agentic AI crushes human costs in L1/L2 support with 85% savings, 300% efficiency gains, and 90%+ success rates.
Your CFO just asked: "Why spend $2.3M annually on L1 support when competitors automated 80% for under $200K?" The humans vs agentic AI cost debate isn't theoretical - it's happening now.
Real Cost Breakdown:
- India L1/L2 Agent: ₹6,64,379 ($8,000) annually (full cost)
- US L1/L2 Agent: $70,000 annually (loaded cost)
- Qcall.ai Agentic AI: ₹6/min ($0.07/min) = 85% savings
Game-Changing Results:
- 47 human agents → 8 agents + AI system
- Operating costs: $2.8M → $420K (85% reduction)
- Customer satisfaction: 78% → 91%
- First call resolution: 72% → 94%
- Languages: 2-3 → 15+ fluently
Why Agentic AI Wins:
- Human Limitations: 8hrs/day, 1-2 languages, 30-45% attrition, mood-dependent, 6-12 weeks training
- Qcall.ai Advantages: 24/7 availability, 97% humanized voice, <60 second response, intelligent qualification, seamless handoff
Industry Impact:
- E-commerce: 82% cost reduction
- Healthcare: 90% appointment automation
- SaaS: 23% satisfaction improvement
- Financial: 95% inquiry automation
Implementation:
- Setup: 2-5 days vs 6-12 weeks hiring
- ROI: Break-even in 3 months
- Scaling: Instant global expansion
Competitive Gap:
- Early Adopters: 85% cost advantage, superior metrics, faster expansion
Late Adopters: Higher costs, limited scalability, compounding disadvantage
Similar automation patterns emerge across industries.
Marketing teams use autoposting.ai for social media automation - same human-AI collaboration model driving L1/L2 transformation.
"We reduced support costs by 78% while improving satisfaction by 31%. ROI was immediate and keeps compounding." - COO, Global SaaS Company
Getting Started:
Calculate current L1/L2 costs
Assess repetitive tasks
Request Qcall.ai demo
Plan 20% pilot
Bottom Line: Companies avoiding this shift hemorrhage millions while competitors scale globally with fraction of costs. The question isn't whether agentic AI will replace routine tasks - it's whether you'll implement before competitors gain insurmountable advantage.
Ready to transform operations? This reveals complete cost analysis, implementation strategy, and competitive advantages.
Decipher SEO Solutions for your startup needs.
Decipher SEO Solutions for your startup needs.mathai2 A solution deck that gives you an idea of how you can use Decipher SEO to target keywords, build authority and generate high ranking content.
With features like images to product you can create a E-commerce pipeline that is optimized to help your store rank.
With integrations with shopify, woocommerce and wordpress theres a seamless way get your content to your website or storefront.
View more at decipherseo.com
Key Challenges in Troubleshooting Customer On-Premise Applications
Key Challenges in Troubleshooting Customer On-Premise ApplicationsTier1 app When applications are deployed in customer-managed environments, resolving performance issues becomes a hidden battle. Limited access to logs, delayed responses, and vague problem descriptions make root cause analysis incredibly challenging.
In this presentation, we walk through:
The six most common hurdles faced in on-prem troubleshooting
A practical 360° artifact collection strategy using the open-source yc-360 script
Real-world case studies covering transaction timeouts, CPU spikes, and intermittent HTTP 502 errors
Benefits of structured data capture in reducing investigation time and improving communication with customers
✅ Whether you're supporting enterprise clients or building tools for distributed environments, this deck offers a practical roadmap to make on-prem issue resolution faster and less frustrating.
Best Practice for LLM Serving in the Cloud
Best Practice for LLM Serving in the CloudAlluxio, Inc. Alluxio Webinar
June 17, 2025
For more Alluxio Events: https://www.alluxio.io/events/
Speaker:
Nilesh Agarwal (Co-founder & CTO @ Inferless)
Nilesh Agarwal, co-founder & CTO at Inferless, shares insights on accelerating LLM inference in the cloud using Alluxio, tackling key bottlenecks like slow model weight loading from S3 and lengthy container startup time. Inferless uses Alluxio as a three-tier cache system that dramatically cuts model load time by 10x.
Building Geospatial Data Warehouse for GIS by GIS with FME
Building Geospatial Data Warehouse for GIS by GIS with FMESafe Software Data warehouses are often considered the backbone of data-driven decision-making. However, traditional implementations can struggle to meet the iterative and dynamic demands of GIS operations, particularly for Iterative Cross-Division Data Integration (ICDI) processes. At WVDOT, we faced this challenge head-on by designing a Geospatial Data Warehouse (GDW) tailored to the needs of GIS. In this talk, we’ll share how WVDOT developed a metadata-driven, SESSI-based (Simplicity, Expandability, Spatiotemporality, Scalability, Inclusivity) GDW that transforms how we manage and leverage spatial data. Key highlights include: - Simplifying data management with a lean schema structure. - Supporting ad hoc source data variations using JSON for flexibility. - Scaling for performance with RDBMS partitioning and materialized views. - Empowering users with FME Flow-based self-service metadata tools.We’ll showcase real-world use cases, such as pavement condition data and LRS-based reporting, and discuss how this approach is elevating GIS as an integrative framework across divisions. Attendees will walk away with actionable insights on designing GIS-centric data systems that combine the best of GIS and data warehouse methodologies.
Advance Doctor Appointment Booking App With Online Payment
Advance Doctor Appointment Booking App With Online PaymentAxisTechnolabs https://www.slideshare.net/slideshow/advance-doctor-appointment-booking-app-with-online-payment/280654375
Simplify Insurance Regulations with Compliance Management Software
Simplify Insurance Regulations with Compliance Management SoftwareInsurance Tech Services Streamline your compliance processes and stay up-to-date with changing regulations effortlessly. Our Compliance Management Software helps insurers reduce risk, avoid penalties, and maintain full compliance — all from a unified platform. Explore More - https://www.damcogroup.com/insurance/compliance-management-software
from ai.backend import python @ pycontw2018
- 1. Joe | AI Engineer
- 2. from ai.backend import python A (backend → AI) story and several thoughts about the future
- 3. Joe Tseng • Backend (AI) engineer • Tainan.py - organizer • PyConTW 2015/2016 - speaker
- 4. Outline About the story Umbo Computer Vision intro Develop and maintain CV services Build our machine learning pipelines What I have learned About the future My imagination
- 5. Outline About the story ➡ Umbo Computer Vision intro Develop and maintain CV services Build our machine learning pipelines What I have learned About the future My imagination
- 6. Autonomous video security systems
- 7. 400 million cameras ref: UK Police Scientific and Development Branch (PSDB) 12 or less cameras per operator - UK Police Scientific Department
- 8. Our Solution Existing cameras Umbo Learning Cameras Real-time alerts Umbo Light
- 9. Infra We use a lot of cloud services and have our own GPU clusters
- 10. Backend Languages All our AI team members speak Python ❤ …
- 11. Flow ./enter_dev_env.sh + write code with unit tests + code review + CI/CD
- 12. Monitoring We log, monitor and measure our services
- 13. Teams Hardware, Dev, CV, Infra, Research, PoC, QA, FAE …
- 14. Develop and maintain all computer vision and machine learning features The definition of an AI Engineer
- 15. 1. Develop and maintain CV services 2. Build our machine learning pipelines Our jobs
- 16. Backend → AI How ?
- 17. Outline About the story Umbo Computer Vision intro ➡ Develop and maintain CV services Build our machine learning pipelines What I have learned About the future My imagination
- 18. • On production in 2016 • From PoC projects to online services Previously on SmartCloud and Umbo Light
- 19. The Simplified CV Architecture Media Server CV Forwarder Load Balancer ZMQ Stream Router Stream Manager Stream Manager Stream Manager Load Balancer CV Worker CV Worker CV Worker HTTP Alert Endpoint
- 20. Where is AI?
- 23. How much can the backend do?
- 24. 87% How much the backend can do
- 25. • Infra • LB / ASG / VPC … • Monitoring • PaperTrail, Cloud Watch … • Development • Add new features • CI/CD • Maintenance • Upgrade packages • Refactor • Debug 87% MS CF LB SR SM SM SM LB CW CW CW AE
- 26. • Infra • LB / ASG / VPC … • Monitoring • PaperTrail, Cloud Watch … • Development • Add new features • CI/CD • Maintenance • Upgrade packages • Refactor • Debug 87% MS CF LB SR SM SM SM LB CW CW CW AE
- 27. • Stream Router/Manager was written by Python2 ❤ ❤ • Python2 => Python3 • byte string / decode • isinstance • format • API changes • Results • Python3 ❤ ❤ ❤ • Enable a lot of cool features • F-strings, typing, asyncio, tracemalloc, etc. 87%
- 28. • Refactor Stream Manager • Happy to use async/await • Use pipeline pattern • Still use run_in_executor/ ThreadPoolExecutor if needed • Results • Cleaner architecture • Performance boost 87%
- 29. • Refactor CV Worker • Lua => Python • Torch => PyTorch • Use aiohttp server • Results • Easier to maintain/upgrade • High performance • GPU resources are the bottleneck • Python package ecosystem 👍 87%
- 30. • Debug • Use pyflame to profile the program • Use tracemaclloc to find memory usage • Use Valgrind to check your C++ code • Results • Understand more and gotcha! 87%
- 31. Ping me if you want to listen this story (and also give us a talk) in Tainan.py 😼
- 32. 13% How much the backend cannot handle
- 33. • About Machine Learning • Not so easy to learn 13%
- 34. More than 13% In fact
- 35. • Computer Vision and Machine Learning • They are everywhere • The learning curve is steep More than 13% MS CF LB SR SM SM SM LB CW CW CW AE
- 36. You need to delve really deep into Computer Vision and Machine Learning to build a self-improving system
- 37. Outline About the story Umbo Computer Vision intro Develop and maintain CV services ➡ Build our machine learning pipelines What I have learned About the future My imagination
- 38. Why machine learning pipelines? Improve services => customers 😆
- 39. How to improve your services Measure first New models / heuristics approaches Need to do training or redesigning Need more (or specific) labeled data Training! Need to compare with the existing ones Get the best on production … Measure again 👷
- 41. • Meet Michelangelo: Uber’s Machine Learning Platform • Machine Learning Pipeline for Real- Time Forecasting @Uber Marketplace Michelangelo (Uber)
- 42. How much can backend do?
- 43. 0%~87%
- 44. Why 0%? 🐵 without domain knowledge
- 45. Why 87%? With domain knowledge + good communication with others + speaking the same language (Python!)
- 46. Catch a glimpse of our pipelines
- 47. • Can dispatch tasks to • Our own labelers • Third-party partners • MTurkers • Backend: • Flask + plugins • NumPy • Celery/Redis/Pymongo • uWSGI Labeling Platform source: https://github.com/nightrome/cocostuff
- 48. • Can train you a lot of new models with different parameters • Backend: • Airflow • Celery Training Pipeline
- 49. • Can visualize • The ground truth data • The evaluation results • Backend: • Flask + plugins • NumPy • SQLAlchemy Visualization Platform
- 50. A faster pipeline means a shorter turnaround time! Speed matters for a startup
- 51. Outline About the story Umbo Computer Vision intro Develop and maintain CV services Build our machine learning pipelines ➡ What I have learned About the future My imagination
- 54. Domain Knowledge Have a basic understanding of ML/CV with practical experience
- 56. Outline About the story Umbo Computer Vision intro Develop and maintain CV services Build our machine learning pipelines What I have learned About the future ➡ My imagination
- 57. from ai.backend import * Python, Golang, Java, C#, Javascript, PHP, Ruby, etc.
- 58. To build your pipelines, Python is the best default language Pipeline → Service
- 59. Every young engineer knows machine learning But some of them may have not heard of GIL 😥
- 60. High-quality (ai) backends are needed
- 61. Thank you
- 62. • Backend, CV, Front-end, App, QA, FAE, Firmware • [email protected] We are hiring