LTIM Recommendations

Syllabus for MTech (AI & DS)

Document Version / Details: Ver. 1.5 / 12-06-2024

 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Course Overview
Today’s modern world is majorly driven Artificial Intelligence (AI) which has made inroads almost
every aspect of life. It was way back in 1959 Arthur Samuel, a pioneer in the field of machine learning
(ML) defined it as the “field of study that gives computers the ability to learn without being explicitly
programmed”. Machine Learning is one of the fundamental blocks of AI. The term "AI" could be
attributed to John McCarthy of MIT (Massachusetts Institute of Technology), which Marvin Minsky
(Carnegie-Mellon University) defines as "the construction of computer programs that engage in tasks
that are currently more satisfactorily performed by human beings because they require high-level
mental processes such as: perceptual learning, memory organization and critical reasoning.

AI has come a long way since then. It was during the late 2000 era that we saw the rebirth of neural
networks in the form of Deep Learning, thanks to the availability of large compute power through
cloud and massive availability of data. Deep Learning is a term to describe deep and complex neural
networks that magically solved at ease the problems around structured and unstructured data (vision
and language). A few years apart came the Generative Adversarial Networks followed by Large
Language Models, now heading towards Large Multimodal Models, Large Action Models. The rise of
AI has been exponential to say the least and is likely to keep going. It’s no wonder, we see robots
cleaning the floor, dev co-pilots generating code across languages for software developers, devices
that human language and follow instructions etc., the list is almost endless. Hence its pertinent that
we are prepared for the future that embraces AI.

We offer M. Tech DS which is a specially designed two-year post-graduate programme for engineers
to leverage the advantages of emerging trends in market in the domain of artificial intelligence and
machine learning and get them skilled to do their job as soon as they complete their Postgraduate.

We focus on building concepts on Mathematics and Statistics required for DS domain, programming,
applied machine learning, generative ai and its applications, advanced deep learning, fine tuning large
language models & building applications and responsible AI. This M. Tech offering is an amalgam of
the fundamentals, essentials of DS and the intricate deep practical knowledge of engineering AI & DS
applications.

The curriculum and syllabus are designed and developed by LTIMindtree Industry practitioners, chief
architects, and experts who have more than 2 decades of experience in this field with rich technical
and domain knowledge, and it will be integrated with a real time lab simulation to help students get
industry exposure while learning.

The outcome of this course is to shape a student’s career in emerging technologies for designing and
implementing the DS applications end-to-end.

Total programme consists of theory, theory + integrated lab, exclusive laboratory, soft skills, and
projects.

2
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Course Structure
PG program M.Tech (AI & DS) Total: 70 credits
Semester I
Course
Subject Hours Type L T P Credits
Code
Mathematical Foundations 60 C 4 0 0 4
Python for Data Science 45 C 3 0 0 3
Applied Machine Learning 45 C 3 0 0 3
Data Engineering 45 C 3 0 0 3
Professional Elective - I 45 C 3 0 0 3
Professional Skills - I 30 P 0 0 2 1
Applied ML Lab 60 P 0 0 4 2
Data Science Lab 60 P 0 0 4 2
Total 390 21
Professional Elective - I: Generative Adversarial Networks (Mandate) / CS62101/ CS60308

Semester II
Course
Subject Hours Type L T P Credits
Code
Deep Learning Applications 45 C 3 0 0 3
Generative AI with Large Language Models 45 C 3 0 0 3
Professional Elective - II 45 C 3 0 0 3
Open Elective 45 C 3 0 0 3
Professional Elective - III 45 E 3 0 0 3
Professional Skills - II 30 P 0 0 2 1
GenAI based Applications Lab 60 P 0 0 4 2
Large Language Models Lab 60 P 0 0 4 2
Total 375 20
Professional Elective - II: Ethics in Data Science (Mandate) / Computer Vision / NLP
Professional Elective - III: Application Architecture and Deployment / Security for Data Science

Semester III
P
Subject Course Code Type L T Credits
(Hours)
Professional Elective - IV C 3 0 0 3
Project - I P - - - 16
Total 19
Professional Elective - IV: Machine Learning Engineering for Production (MLOps) Specialization / Industry Specific Applications of
GenAI & Responsible AI

Semester IV
Subject Course Code Type L T P Credits
Project -II P - - - 20
Total 20

C-Core, E-Elective, A- Audit, O- Open elective, P- Practical, L- Lecture, T- Tutorial

3
©LTIMindtree | Privileged and Confidential 2024
LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Semester - I

4
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Title Mathematical Foundations Code

Prerequisite Credits | Total Hours 4-0-0 [4] | 60

Course Outcome:
Students will be able to refresh the statistical knowledge learnt earlier with hands-on practical expertise.

CO1: Refresh the mathematics knowledge with respect to Linear algebra, Vectors, Projections, Principal Component
Analysis and Generative Models
CO2: Refresh the mathematics knowledge with respect to Matrices, Gradient Calculus, Optimization models.
CO3: Refresh the mathematics knowledge with respect to probability, statistics.

Unit 1: Linear Algebra 12

Systems of Linear Equations - Machine learning motivation - A geometric notion of singularity - Singular vs non-singular
matrices - Linear dependence and independence - Matrix row-reduction - Row operations that preserve singularity -
The rank of a matrix - Row echelon form - Reduced row echelon form- LU decomposition- Solving Systems of Linear
Equations - Machine learning motivation - Solving non- singular systems of linear equations - Solving singular systems
of linear equations - Solving systems of equations with more variables - Gaussian elimination.

Unit 2: Probability & Statistics 12

Introduction to probability - Concept of probability: repeated random trials - Conditional probability and independence
- Random variables - Cumulative distribution function - Discrete random variables: Binomial distribution - Probability
mass function - Continuous random variables: Uniform distribution - Continuous random variables: Gaussian
distribution -Joint distributions - Marginal and conditional distributions - Independence - covariance - Multivariate
normal distribution - Sampling and point estimates - Interval estimation -Confidence intervals – Confidence Interval
for mean of population - Biased vs Unbiased estimates-Maximum likelihood estimation - Intuition behind maximum
likelihood estimation - Hypothesis testing - Describing samples: sample proportion and sample mean - Two types of
errors - Test for proportion and means - Two sample inference for difference between groups.

Unit 3: Bayesian Statistics & its applications in various fields 12

Bayesian statistics and its applications in various fields - Bayesian Learning: Bayes theorem - maximum likelihood and
least squared error hypotheses – Naïve Bayes classifier- Bayesian belief networks- gradient ascent training of Bayesian
networks- learning the structure of Bayesian networks- the EM algorithm- mixture of models- Markov models- hidden
Markov models - Time series analysis and forecasting techniques - Basic Properties of time-series data: Distribution
and moments- Stationarity- Autocorrelation- Heteroscedasticity- Normality- Survival Analysis.

Unit 4: Non-Parametric Statistics 12

Non-parametric Statistics - Chi square test- Sign test -Wilcoxon signed rank test - Mann Whitney test - Run test -
Kolmogorov Smirnov test - Spearmann and Kendall’s test - Tolerance region.

Unit 5: Multivariate Statistical Methods for Analyzing Complex Datasets 12

Multivariate statistical methods for analysing complex datasets - Factor Analysis - Cluster Analysis- Regression Analysis
- Discriminant Analysis.

5
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Reference Books:
1. James D. Miller, Statistics for Data Science - By Packt Publishing (2017)
2. IND James D. Hamilton, Time Series Analysis – By Levant Books (2012)

Title Python for Data Science Code

Prerequisite Credits |Total Hours 3-0-0 [3] | 45

Course Outcome:
Students will be able to learn about two widely used programming languages in the field of data science and how to
go about choosing any language.

CO1: Understand the basics of python and standard modules used for data science with hands-on.
CO2: Understand the data structures and visualization used for data science with hands-on.
CO3: Understand the machine learning libraries used for data science with hands-on.

Unit 1: Python - Data Structures, OOPS & Modules 9

Data structures: Dictionaries - Maps - Hash Tables - Array Data Structures - Records - Structs - Data Transfer Objects -
Sets and Multisets-Stacks (LIFOs) - Queues (FIFOs) ; Python : Python installation - Python OOPs - Polymorphism in OOPs
programming - Python String Concatenation - Print Exception in Python - Python Libraries - Python Pandas - Python
Matplotlib - Python Seaborn - Python SciPy - Chatbot in Python - Machine Learning using Python - Exploratory Data
Analysis in Python - Open CV Python - Tkinter - Pythons Turtle Module - PyGame in Python - Pytorch - Scrapy - Web
Scraping - Django - Python Programs - Types of Data structure in Python - Built in data structures - User defined data
structures; Object Oriented Concepts and Design : APIs and Data Collection - Simple API - REST APIs & HTTP Requests
- Web scraping - HTML for Web Scraping - file formats

Unit 2: Python – Numpy, Pandas & DS Libraries 9

Installation and setup : Anaconda Distribution - Anaconda Navigator to create a New Environment - Startup and
Shutdown Process - Intro to the Jupyter Lab Interface - Code Cell - execution; Python : Basic datatypes - Operators -
variables - Built in Functions - Custom Functions - String Methods - Lists - Index Positions and Slicing - Navigating
Libraries using Jupyter Lab; Series : Create series object from a list and dictionary - The head and Tail methods - Passing
Series to Python Built-In Functions – Methods for Data sorting ; Dataframe : Methods and Attributes between Series
and DataFrames - Fill in Missing Values - Filtering data and methods in Dataframe - Data Extraction in dataframes -
Working with Text Data - Merging Dataframes; Data Mining - Data Processing and Modelling - Data Visualization

Unit 3: Visualization 9
Introduction to Matplotlib - Matplotlib Basics - Matplotlib - Understanding the Figure Object - Matplotlib -
Implementing Figures and Axes - Matplotlib - Figure Parameters - Matplotlib Styling - Legends - Matplotlib Styling -
Colors and Styles - Advanced Matplotlib Commands - Introduction to Seaborn - Scatterplots with Seaborn - Distribution
Plots - Part One - Understanding Plot Types - Distribution Plots - Part Two - Coding with Seaborn - Categorical Plots -
Statistics within Categories - Understanding Plot Types - Categorical Plots - Statistics within Categories - Coding with
Seaborn - Categorical Plots - Distributions within Categories - Understanding Plot Types - Categorical Plots -
Distributions within Categories - Coding with Seaborn - Seaborn - Comparison Plots - Understanding the Plot Types -
Seaborn - Comparison Plots - Coding with Seaborn - Seaborn Grid Plots - Seaborn - Matrix Plots.

6
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Unit 4: Regression and Classification 9

Introduction to Linear Regression : Cost Functions - Gradient Descent - Python coding Simple - Overview of Scikit-Learn
and Python - Residual Plots - Model Deployment and Coefficient Interpretation - Polynomial Regression - Theory and
Motivation - Creating Polynomial Features - Training and Evaluation - Bias Variance Trade-Off - Polynomial Regression
- Choosing Degree of Polynomial - Model Deployment - Feature Scaling; Introduction to Cross Validation :
Regularization Data Setup - Ridge Regression Theory - Lasso Regression - Background and Implementation - Elastic Net
- Feature Engineering and Data Preparation; Dealing with Outliers - Dealing with Missing Data - Evaluation of Missing
Data - Filling or Dropping data based on Rows - Fixing data based on Columns - Dealing with Categorical Data - Encoding
Options - Cross Validation - Test - Validation - Train Split - cross_val_score - cross validate - Grid Search; Linear
Regression Project: The Logistic Function - Logistic Regression - Theory and Intuition; Linear to Logistic: Logistic
Regression - Theory and Intuition - Linear to Logistic Math; Logistic Regression: Theory and Intuition Logistic Regression
Model Training - Classification Metrics - Confusion Matrix and Accuracy - Classification Metrics - Precison, Recall, F1-
Score - ROC Curves - Logistic Regression with Scikit-Learn - Performance Evaluation - Multi-Class Classification with
Logistic Regression - Data and EDA – Model.

Unit 5: Unsupervised and Advanced Machine Learning 9

Introduction to KNN Section: KNN Classification, KNN Coding with Python - Choosing K, KNN Classification Project
Exercise; Introduction & history of Support Vector Machines- Hyperplanes and Margins, Kernel Intuition, Kernel Trick
and Mathematics; SVM with Scikit-Learn and Python – Classification, Regression Tasks; Introduction to Tree Based
Methods- Decision Tree, Understanding Gini Impurity; Constructing Decision Trees with Gini Impurity, Coding Decision
Trees; Introduction to Random Forests-Key Hyperparameters, Number of Estimators and Features in Subsets,
Bootstrapping and Out-of-Bag Error; Coding Classification with Random Forest Classifier, Coding Regression with
Random Forest Regressor, Advanced Models. Introduction to K-Means Clustering Section; K-Means Color Quantization;
K-Means Clustering Exercise Overview, Solution; Introduction to Hierarchical Clustering, Coding - Data and
Visualization, Scikit-Learn; Introduction to Principal Component Analysis(PCA)-Manual Implementation in Python-
SciKit-Learn.

Reference Books:

1. Alvaro Fuentes, Become a Python Data Analyst – By Packt Publishing (2018)

2. Bharti Motwani, Data Analytics using Python – By Wiley (2020)
3. Jules S. Damji, Learning Spark: Lightning-Fast Data Analytics, Second Edition – By Shroff/O'Reilly (2020)

Title Applied Machine Learning Code

Prerequisite Credits |Total Hours 3-0-0 [3] | 45

Course Outcome:
Students will be able to get the knowledge about all the tools and techniques you need to apply machine learning to
solve business problems.

CO1: To know about Supervised Learning, Support Vector Machines, Unsupervised Learning
CO2: Get the knowledge about Feature Engineering, Statistical Data Analysis, Outlier Analysis and Detection
CO3: Learn about ML Model Development, Model Evaluation Techniques, Model Deployment and Inferences, Model
Explainability

Unit 1: Supervised Learning 15

7
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Implement and understand the cost function and gradient descent for multiple linear regression - Implement and
understand methods for improving machine learning models by choosing the learning rate - plotting the learning curve
- performing feature engineering - applying polynomial regression - Implement and understand the logistic regression
model for classification -Learn why logistic regression is better suited for classification tasks than the linear regression
model is - Implement and understand the cost function and gradient descent for logistic regression - Understand the
problem of - overfitting - improve model performance using regularization - Implement regularization to improve both
regression and classification models

Unit 2: Advanced Learning Algos 15

Build a neural network for binary classification of handwritten digits using TensorFlow - Gain a deeper understanding
by implementing a neural network in Python from scratch - Optionally learn how neural network computations are
vectorized to use parallel processing for faster training and prediction - Build a neural network to perform multi-class
classification of handwritten digits in TensorFlow -using categorical cross-entropy loss functions and the SoftMax
activation - Learn where to use different activation functions – ReLu - linear - sigmoid - SoftMax in a neural network -
depending on the task you want your model to perform - Use the advanced Adam optimizer to train your model more
efficiently - Discover the value of separating your data set into training - cross-validation -test sets - Choose from
various versions of your model using a cross-validation dataset -evaluate its ability to generalize to real-world data
using a test dataset - Use learning curves to determine if your model is experiencing high bias or high variance - learn
which techniques to apply regularization - adding more data - adding or removing input features to improve your
model’s performance - Learn how the bias-variance trade-off is different in the age of deep learning - and apply Andrew
Ng’s advice for handling bias and variance when training neural networks - Learn to apply the iterative loop of machine
learning development to train - evaluate - tune your model - Apply data-centric AI to not only tune your model but
tune your data using data synthesis or data augmentation to improve your model’s performance - Build decision trees
and tree ensembles - such as random forest and XGBoost - boosted trees - to make predictions - Learn when to use
neural network or tree ensemble models for your task - as these are the two most commonly used supervised learning
models in practice today.

Unit 3: Unsupervised Learning and Recommender Systems 15

Use unsupervised learning techniques for unsupervised learning: including clustering and anomaly detection - Build
recommender systems with a collaborative filtering approach and a content-based deep learning method - Build a
deep reinforcement learning model - Implement K-mean clustering - Implement anomaly detection - Learn how to
choose between supervised learning or anomaly detection to solve certain tasks - Build a recommender system using
collaborative filtering - Build a recommender system using a content-based deep learning method - Build a deep
reinforcement learning model (Deep Q Network)." - Histograms - Box Plots etc - use of frequency distributions – mean
comparisons - cross tabulation - statistical inferences using chi square - t-test and ANOVA - Outlier Analysis and
Detection - outlier analysis - density based and distance based.

Reference Books:
1. Hang Li, Machine Learning Methods - By Springer Nature Singapore (2023)
2. Dr. R. Nageswara Rao, Machine Learning in Data Science Using Python - By Dreamtech Press (2022)

8
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Title Data Engineering Code

Prerequisite Credits |Total Hours 3-0-0 [3] | 45

Course Outcome:
Students to understand the fundamentals of data engineering and its importance in modern data-driven applications.

CO1: Identify and explain different data formats and their use cases, including structured, semi-structured, and
unstructured data.
CO2: Describe various data ingestion techniques, such as ETL, and stream processing, and their advantages and
limitations.
CO3: Perform data profiling and analyze data quality metrics to ensure data accuracy, completeness, and consistency.
CO4: Design and implement effective storage and retrieval methods for large-scale data sets, including relational
databases, NoSQL databases, and distributed file systems.
CO5: Apply data engineering principles to real-world scenarios, such as data warehousing, big data analytics, and
machine learning.

Unit 1: Data Types & Formats 9

Introduction to Data Types and Formats - Types of Data - Structured vs. Unstructured Data - Formats of Data - Semi-
Structured Data - Data Type Conversion and Transformation - Data Serialization - Choosing the Right Data Type and
Format - Tools and Technologies for Data Types and Formats.

Unit 2: Data Ingestion Techniques 9

Introduction to Data Ingestion - Streaming Data Ingestion - Batch Data Ingestion - Hybrid Data Ingestion - Data Ingestion
vs. Data Integration - Data Ingestion Challenges - Tools and Solutions for Data Ingestion - StreamSets DataOps Platform
- Benefits of Data Ingestion - Data Ingestion Framework.

Unit 3: Data Profiling & Visual Representation via Various Tools (Pandas) 9
Introduction to Data Profiling and Visualization - Exploratory Data Analysis (EDA) with Pandas - Steps Involved in
Exploratory Data Analysis (EDA) Data Analysis (EDA) with Pandas - Market Analysis with Exploratory Data Analysis (EDA)
- Data Analytics and Its Future Scope - Data Analytics with Python - Top Business Intelligence Tools - Application of Data
Analytics - Retrieving and Cleaning Data - Exploratory Data Analysis and Feature Engineering - Inferential Statistics and
Hypothesis Testing - Descriptive Statistics - Types of Descriptive Statistics - Concepts of Populations, Samples, and
Variables - Statistical Methods for Describing Data Characteristics - Real-World Applications of Descriptive Statistics
using Excel - Types of Missing Data and Handling Techniques.

Unit 4: Storage and Retrieval Methods 9

Introduction to Storage and Retrieval - Types of Data and Storage Methods - Local vs. Distributed Storage & Retrieval
- Hardware Aspects of Storage & Retrieval - Choosing Storage Methods - Data Partitioning and Sharding - Data
Replication and Redundancy - Data Compression and Encoding - Data Archiving and Retrieval - Backup and Disaster
Recovery - Data Lifecycle Management.

Unit 5: Data Lineage Analysis 9

9
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Introduction to Data Lineage Analysis - Building a Data Flow - ETL (Extract, Transform, Load) Process - Usage of Data
Warehouse - Edge Intelligence in Data Flow - Understanding Data Lineage - How Data Lineage Works - Benefits of Data
Lineage - Data Lineage Tool Features.

Reference Books:
1. Charles M.Judd, Data Analysis: A Model Comparison Approach To Regression, ANOVA, and Beyond 3rd
Edition - By Routledge (2017)
2. Pierre-Yves Bonnefoy, Emeric Chaize, Raphaël Mansuy & Mehdi TAZI, The Definitive Guide to Data
Integration 1st Edition - By Packt Publishing (2024).

Title Generative Adversarial Networks (Professional Elective – I) Code

Prerequisite Credits |Total Hours 3-0-0 (3) | 45

Course Outcome:
Students gain comprehensive understanding of deep learning techniques and generative AI models.

CO1: Understand generative models such as generative adversarial networks (GANs) and their advanced techniques.
CO2: To build sophisticated and robust GAN models using PyTorch & convolutional layers etc.,
CO3: Student to learn about the advantages and disadvantages of different GAN performance measures.
CO4: Students to explore and examine the applications of GANs

Unit 1: Build Basic Generative Adversarial Networks (GANs) 15

Overview of GenAI - Intro to GANs - Learn about GANs and their applications, understand the intuition behind the basic
components of GANs -build your very own GAN using PyTorch - Deep Convolutional GAN - Build a more sophisticated
GAN using convolutional layers - Learn about useful activation functions - batch normalization - and transposed
convolutions to tune your GAN architecture and apply them to build an advanced DCGAN specifically for processing
images - Wasserstein GANs with Normalization - Reduce instances of GANs failure due to imbalances between the
generator and discriminator by learning advanced techniques such as WGANs to mitigate unstable training and mode
collapse with a W-Loss and an understanding of Lipschitz Continuity - Conditional and Controllable GANs - Understand
how to effectively control your GAN - modify the features in a generated image - and build conditional GANs capable
of generating examples from determined categories.

Unit 2: Build Better Generative Adversarial Networks (GANs) 15

GAN Evaluation - Understand the challenges of evaluating GANs - learn about the advantages and disadvantages of
different GAN performance measures - and implement the Fréchet Inception Distance FID method using embeddings
to assess the accuracy of GANs -GAN Disadvantages and Bias - Find out the disadvantages of GANs when compared to
other generative models - discover the pros/cons of these models — plus - learn about the many places where bias in
machine learning can come from - why it’s important - and an approach to identify it in GANs - StyleGAN and
Advancements - Understand how StyleGAN improves upon previous models and implements the components and the
techniques associated with StyleGAN - currently the most state-of-the-art GAN with powerful capabilities.

Unit 3: Apply Generative Adversarial Networks (GANs) 15

GANs for Data Augmentation and Privacy Preservation - Explore the applications of GANs and examine them wrt data
augmentation, privacy, and anonymity Improve your downstream AI models with GAN-generated data - Image-to-

10
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Image Translation - Leverage the image-to-image translation framework and identify extensions – generalizations -
applications of this framework to modalities beyond images -Implement Pix2Pix - a paired image-to-image translation
GAN - to adapt satellite images to map routes with advanced U-Net generator -Patch GAN discriminator architectures
- Image-to-Image Unpaired Translation - Compare paired image-to-image translation to unpaired image-to-image
translation and identify how their key difference necessitates different GAN architectures - Implement CycleGAN- an
unpaired image-to-image translation model, to adapt horses to zebras with two GANs in one.

Reference Books:
1. Jakub Langr & Vladimir Bok, GANs in Action: Deep learning with Generative Adversarial Networks - By
Manning; 1st edition (2019)
2. John Hany, Hands-On Generative Adversarial Networks with PyTorch 1.x - By Packt Publishing, (2019)

Title Professional skill - I Code

Prerequisite Credits |Total Hours 0-0-2 [1] | 30

Course Outcome:
Students to understand professional, behavioral and presentation skills while working with team and practically
experience important aspects of it.

CO1: To help the students understand and implement positive outlook, interpret the body language of team members
and stakeholders, better interpersonal relationships. Develop into self-motivated professionals with confidence.
Practice Responding instead of Reacting.
CO2: Create good Presentation and Present with confidence. Also, recognize and manage Stress, Prioritize and Plan.
CO3: To listen to understand. To be able to ask good questions.
CO4: To understand to be a good Team player, Team Dynamics and to understand the Business Ethics
CO5: To be able to write and speak correctly, forming grammatically correct sentences.

Unit 1: Positive Attitude 6

Attitude- Campus to Corporate attitude change, Recognizing Negative Attitude, Campus to Corporate attitude change;
Attitude at work- Impact of Negative Attitude in the Workplace, Overcoming Negative Attitude, positive attitude,
thought process, Building self-confidence and Assertiveness; Toxic positivity; 3Es, Motivation-Intrinsic and Extrinsic
Motivation, Inspiration vs motivation; Emotional Intelligence-Intro to EI, Four clusters. Transactional Analysis (TA),
SWOT analysis - Professional analysis

Unit 2: Body Language 6

Importance of Body Language, Five Cs of Body Language, Body language in different cultures, Positive Body Language;
Voice Control- Pace. Pause and Pitch; Culture-Inclusivity and Proxemics across Global Cultures, Understanding POSH;
Stress Management-What is Stress, Eustress, Reasons of stress (work/ personal); Stress Management Techniques

Unit 3: Presentation Skills 6

Self-introduction – Exercises, Why Give Presentations; Craft your message-Plan the visuals, Manage the Response;
How to create an effective presentation - Virtual & Physical, Do’s & Don'ts of Presentation Skills, Objection handling,
Stage Fear – Causes and Cure, Practice the Delivery; Time Management-Common Time & Energy Wasters, Planning &
Prioritizing Time Matrix & Analysis

11
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Unit 4: Listening & Questioning skills 6

Barriers to effective listening - how to overcome them; Exercises - Customer Call Flow – Role-play, Cust calls amongst
the team; How to frame Questions, Different kinds of questions, asking appropriate questions; Spoken English-
Introduction to Parts of Speech and its usage; Subject - Verb Agreement; Basic conversation skills-sentence
construction -SVO

Unit 5: Teamwork 6
Teamwork and Ethics - Definition of TEAM - Team vs Groups. Difference b/w Healthy competition and cut throat
competition, Importance of working in teams, Evolution of a TEAM, Benefits of team work; Virtual teams- Challenges
and ways to overcome it, Diversity and Inclusion in a team; Development of Teams Stages of team development; Team
dynamics-its importance & Interpersonal Skills Development Ethics- to enable students to identify and deal with ethical
problems, develop their moral intuitions, which are implicit in everyday choices and actions; Conflict Management:
Team building Activities- Predetermined/ Predesigned Indoor/ Outdoor activities to build a team, enhance language
and inter personal skills

12
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Applied Machine Learning Lab

Experiment No Experiment Details Type
Understanding "Mobile Price" dataset by doing feature analysis. Data is available at:
Experiment 1 Individual
https://www.kaggle.com/datasets/iabhishekofficial/mobile-price-classification/data
Execute data preprocessing step on the above dataset: perform outlier and missing data
Experiment 2 Individual
analysis towards building a refined dataset
Build machine learning model/s to predict the actual price of the new mobile based on
Experiment 3 Individual
other given features like RAM, Internal Memory etc
Calculate the prediction accuracy of the models used in Experiment 3 and do comparative
Experiment 4 Individual
analysis among them to identify the best technique.
Understanding "Second Hand Car Prediction Price" dataset by doing feature analysis. Data
Experiment 5 is available at: https://www.kaggle.com/datasets/sujithmandala/second-hand-car-price- Individual
prediction
Perform data preprocessing step on the above dataset: perform outlier and missing data
Experiment 6 Individual
analysis towards building a refined dataset.
Perform Feature Engineering towards building new feature which is more impactful.
Experiment 7 Build machine learning model/s to predict the price of the car based on other given Individual
features like Brand, Model, Year, Fuel Type etc
Calculate the prediction accuracy of the models used in Experiment 7 and do comparative
Experiment 8 Individual
analysis among them to identify the best technique.
Plot the features (actual price and predicted price) in scatter plot to understand the
Experiment 9 Individual
variation.
Understanding "Marketing Campaign Positive Response Prediction" dataset by analysing
all the features. Data is available at:
Experiment 10 Individual
https://www.kaggle.com/datasets/sujithmandala/marketing-campaign-positive-
response-prediction
Perform exploratory data analysis on the above dataset: perform outlier and missing data
Experiment 11 analysis towards building a refined dataset. Show the outliers in box plot or through some Individual
statistical technique. Find the numerical and categorial features.
Perform Feature Engineering towards building new feature which is more impactful than
Experiment 12 the existing ones. Build the correlation matrix and show visually the relationship among Individual
various features.
Build machine learning model/s to predict the result of marketing campaign based on
Experiment 13 Individual
other given features like customer details, gender, annual income etc
Calculate the prediction accuracy of the models used in Experiment 13 and do comparative
Experiment 14 Individual
analysis among them to identify the best technique.
Please check whether you find imbalanced classes, overfitting, and data bias in the above
Experiment 15 Individual
two datasets. Please apply some technique to overcome it.

13
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Data Science Lab

Experiment No Experiment Details Type
Present your view on the different techniques you have
employed to do outlier analysis, handling missing data,
feature engineering, feature importance and improving
Case Study 1 Group Project
the accuracy of the model both from a classifier as well
as a regressor. Use any sample data and present your
POV in a well-structured presentation.
Present your findings on different activation functions Group Project
you have used and methods to improve the accuracy of
the model using neural networks. You should be able to
Case Study 2
clearly articulate the advantage and disadvantage of
each activation function. Use any sample data and
present your POV in a well-structured presentation.
Present your findings on different techniques of Group Project
anomaly detection and k means clustering. Use any
Case Study 3
sample data and present your POV in a well-structured
presentation
Present your POV on how to generate synthetic data Group Project
using GANs. You can assume a sample dataset from an
Case Study 4
IOT enabled machine where the failure rates are
minimal.
Present your POV on Style related GANS. Explore the Group Project
earliest models to the current models. Articulate the
Case Study 5
successive improvements in the models. Also articulate
the future of GANs in generating realistic images.
Present your POV on GANs used for Deep Fakes. Group Project
Case Study 6 Articulate how we can identify the Deep Fake from the
original.

14
©LTIMindtree | Privileged and Confidential 2024
LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Semester - II

15
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Title Deep Learning Applications Code

Prerequisite Credits |Total Hours 3-0-0 [3] | 45

Course Outcome:
Students learn both theoretical and practical aspects of deep learning applying to build real world applications.

CO1: To understand the fundamentals of deep learning and its applications in computer vision, time series analysis
and natural language processing.
CO2: To build a facial recognition system.
CO3: To build a weather forecasting system.
CO4: To build a chatbot.

Unit 1: Building a Facial Recognition System - Part 1 9

Convolutional Neural Network (CNN): Transfer learning - Data Augmentation - Image segmentation - Object
detection - Video classification - Text and natural language processing - Structured data - Model optimization

Unit 2: Building a Facial Recognition System - Part 2 9

Facial recognition model: Writing the code - Deploying the API as container - Consuming the API from Frontend and
display- Preparing the image dataset - Creating and training the Model; Build and deploy Flask REST API on Docker:
steps to dockerize your flask app; Docker: Docker Installation – Architecture – Working of Docker; Kubernetes:
Overview – Architecture – Kubernetes Setup – Advanced Kubernetes; Flask: Overview – Environment - Application

Unit 3: Building a Facial Recognition System - Part 3 9

Facial Recognition system: Create Endpoints and UI to retrain the system with new data (faces) - Feedback system
for face labels - Transfer Learning - Reusing the knowledge with additional learning; Technology: Flask, streamlit and
Tensorflow - Create a multipage app - API reference - Advanced features - Components - Roadmap - Changelog -
Cheat sheet - Streamlit community cloud.

Unit 4: Building a Weather Forecasting System with Chatbot - Part 1 9

Recurrent Neural Network: Architecture - Technology and libraries - Application of RNN - Limitations of RNN -
Improvement LSTM - RNN in time series - Build an RNN to predict time series in TensorFlow - Text generation with
an RNN ; Chatbot : Working of chatbot - Types of Chatbot - Use cases of chatbots - Objective - End goal - constraints
- How to build a chatbot - A ten - minute introduction to sequence to sequence learning in keras - Chatbot using
seq2seq LSTM models - Architecture of seq2seq model

Unit 5: Building a Weather Forecasting System with Chatbot - Part 2 9

Intelligent Chatbox: Using LSTM - Using NLP - LSTM Time series Analysis - LSTM weather - Create an Intelligent
chatbot in Python using the spaCy NLP Library - Prerequisites - Setting up the environment - creating the city weather
program - Creating the chatbot.

Reading Materials:
1. James D. Hamilton, Time Series Analysis – By Levant Books (2012)

16
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

2. V Kishore Ayyadevara & Yeshwanth Reddy, Modern Computer Vision with PyTorch - By Packt Publishing
(2020)

Title Generative AI with Large Language Models Code

Prerequisite Credits |Total Hours 3-0-0 [3] | 45

Course Outcome:
Students to get the knowledge to adapt pre-trained LLMs to more specialized tasks.

CO1: Understand Fundamentals of Fine Tuning, Types of fine-tuning Techniques.

CO2: Reinforcement learning and LLM-powered applications.

Unit 1: Introduction to Generative AI 15

Introduction Generative AI & LLMs - LLM use cases and tasks - Text generation before transformers - Transformers
architecture - Generating text with transformers - Prompting and prompt engineering (CoT) – RAG Technique for
retrival - Generative configuration - Generative AI project lifecycle - Pre-training large language models -
Computational challenges of training LLMs.

Unit 2: Fine Tuning and Evaluation 15

Instruction fine-tuning - Fine-tuning on a single task - multi-task instruction fine-tuning - Model evaluation –
Benchmarks -Parameter efficient fine-tuning (PEFT) -PEFT techniques 1: LoRA - PEFT techniques 2: Soft prompts.

Unit 3: Reinforcement learning and LLM-powered applications 15

Aligning models with human values - Reinforcement learning from human feedback (RLHF) - RLHF: Obtaining
feedback from humans - Reward model - Fine-tuning with reinforcement learning - Model optimizations for
deployment - Generative AI Project Lifecycle - Using the LLM in applications - Interacting with external applications
- Helping LLMs reason and plan with chain-of-thought - Program-aided language models (PAL) - ReAct: Combining
reasoning and action - LLM application architectures.

Reading Material:

1. Edward R. Deforest, Prompt Engineering with Transformers and LLM – By Kindle (2024).
2. Altaf Rehmani, Generative AI for everyone – By Altaf Rehmani; 1st edition (2024).

Ethics in Data Science (Professional Elective

Title Code
– II)
Prerequisite Credits | Total Hours 3-0-0 [3] | 60

Course Outcomes:
Students to get the knowledge of Ethics in Data Science.

CO1: Understand Philosophical frameworks for assessing fairness.

17
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

CO2: Get knowledge on Data ownership, privacy and anonymity.

Unit 1: Introduction and Philosophical frameworks for assessing fairness 7

Foundations of ethics - early theories of fairness (Utilitarianism etc.) - contemporary theories of fairness - significance
of ethics in data science - ethics vs. law/compliance/public relations - cultural relativism - “professional” ethics in
data science - individuals vs. collectives.

Unit 2: Research Ethics 9

Data driven research, methods of collection of data - different types of data: qualitative and quantitative - overview
of ethical issues in data-driven organizations - doing ethical data analysis - responsible use of research data -
plagiarism - fake data and fabrication of data - creation of data base.

Unit 3: Data ownership, privacy, and anonymity 9

Understanding the difference between data ownership - data privacy and data anonymity - under- standing the idea
behind data surveillance - data privacy vs. data security.

Unit 4: Algorithmic fairness 9

Discrimination and algorithms- obscure and unintentional bias displayed by the algorithms - ethics of data scraping
and storage- Mosaic data- found data- and designed data.

Unit 5: Policies on data protection 9

EU’s general data protection rules - GDPR - digital India policy - personal data protection bill - 2019 -PDP Bill- ethical
issues on data privacy in context with India - case studies.

Unit 6: Responsible AI 8
Various dimensions of Responsible AI - Dimensions of Ethical AI - Bias Mitigation Techniques; Constitutional AI:
Rules of Constitutional AI - How to create Constitutional AI complaint system - Model fine tuning for Constitutional
AI.

Unit 7: Red Teaming on LLM & Case study 9

What are the vulnerabilities - How to attack those problems by Red Teaming.

Natural Language Processing (Professional

Title Code
Elective – II)
Prerequisite Credits | Total Hours 3-0-0 [3] |45

18
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Course Outcome:
Students to understand natural language processing in depth with various factors involved in it.

CO1: Understand the purpose of NLP and how to use it in real world applications with example.
CO2: Understand how to solve a classification problem.
CO3: Understand how deep learning is applied for NLP.
CO4: Understand the transfer learning concepts for reusability of knowledge.
CO5: Understand the applications of voice recognition system.

Unit 1: NLP Need & Real-World Applications 9

What is NLP and its components? - Phases of NLP - Challenges of natural language - Applications of NLP - Industries
using NLP - NLP programming languages - NLP libraries and Development environments - Use of AI in NLP - Basic Text
Processing and Linguistic Concepts: Tokenization - Stemming - Lemmatization - Part-of-Speech Tagging.

Unit 2: Text Classification 9

Benefits of Text Classification - Types of Text classification - Challenges in text classification - Applications of text
classification

Unit 3: Deep Learning for NLP 9

Convolutional Neural Networks (CNNs) for NLP - Recurrent Neural Networks (RNNs) for NLP - Recursive Neural
Networks - Hybrid Models for NLP

Unit 4: Transfer Learning for NLP 9

Benefits of Transfer Learning for NLP - Fine Tuning techniques - Fine-Tune BERT for Spam Classification

Unit 5: Voice Recognition 9

Basics of Voice Recognition: Difference between speech and voice recognition - Use of NLP in voice recognition and
transformation: Speech recognition using NLP models (HMM, DTW) - Acoustic modelling - Error correction in voice
recognition.

Reading Materials:
1. Sowmya Vajjala, Bodhisattwa Majumder, Anuj Gupta & Harshit Surana, Practical Natural Language
Processing – By Shroff/O'Reilly (2020)
2. Uday Kamath, John Liu & James Whitaker, Deep Learning for NLP and Speech Recognition – By Springer
1st ed. (2019)

Computer Vision (Professional

Title Code
Elective – II)

19
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Prerequisite Credits |Total Hours 3-0-0 [3] | 45

Course Outcome:
Students to understand the computer vision techniques in depth with various applications of the same.

CO1: Understand what techniques are available to process the image.

CO2: Understand how to analyze the image and extract required features.
CO3: Understand how computer vision solves real world problems.

Unit 1: Image Processing Techniques 9

Introduction to image processing: What is image processing? - Understanding about types of image processing-
Visualization, Recognition, Sharpening & Restoration, Pattern Recognition, Retrieval; Image Transformation: Image
Enhancement Techniques: Histogram Equalization, Contrast Stretching, Adaptive Enhancement - Image Restoration
Methods: Deblurring, Denoising, Inpainting - Linear Filtering: Convolution, Gaussian Filtering, Edge Detection -
Independent Component Analysis (ICA) - Pixelation and Its Applications; Image Generation Technique: Procedural
Image Generation: Fractal Generation, Noise-based Generation - Generative Adversarial Networks (GANs) for Image
Generation: Introduction to GANs- Understanding the architecture and training process of generative adversarial
networks, Implementing GANs for generating realistic images, including applications in image-to-image translation
and style transfer. - Applications of Image Generation Techniques: Data Augmentation, Creative Applications.

Unit 2: Feature Extraction and Image Analysis 9

Feature Detection: Introduction to feature detection - Object recognition techniques (key point detection, edge
detection) - Image segmentation algorithms (region growing, thresholding, etc.) - Frequency domain processing
(Fourier transform, frequency filtering) - Feature extraction methods (SIFT, SURF); Object Description: Introduction
to fundamentals of moving object detection - Moving object description techniques (optical flow, background
subtraction) - Camera geometry for object description (camera calibration, pose estimation).

Unit 3: Machine Learning for Computer Vision 9

Image Classification: Introduction to machine learning for computer vision - Image classification models (CNNs,
transfer learning) - Object detection with machine learning (YOLO, SSD) - Labeling images for machine learning
(annotation tools, data augmentation).

Unit 4: 3D Computer Vision 9

Depth Perception: Comparison of 2D and 3D computer vision - Real-world applications and trends in 3D computer
vision - Classification of 3D data (point clouds, meshes).

Unit 5: Advanced CV and Future Trends 9

Advanced Computer Vision Applications: Brain Tumor Detection - Integrating Computer Vision in Autonomous
Driving Systems - Computer Vision Applications in the Food Industry;Object Detection and Recognition: Visual
Tracking - Semantic Segmentation - Human Recognition.

Reading Materials:

20
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

1. V Kishore Ayyadevara & Yeshwanth Reddy, Modern Computer Vision with PyTorch - By Packt Publishing
(2020)
2. B Cyganek, An Introduction to 3D Computer Vision Techniques and Algorithms – By John Wiley & Sons Inc;
1st edition (2009)

Title Professional skill - II Code

Prerequisite Credits |Total Hours 0-0-2[1] |30

Course Outcome:
Students understand day in day out terms used in customer environment and demonstrate customer centric
approach and practically experience the and important aspects of it.

CO1: To understand what is spoken without distortion and respond appropriately.

CO2: To behave professionally.
CO3: To participate productively in an official meeting keeping etiquette in mind.
CO4: To communicate effectively through writing.
CO5: To behave appropriately in an official environment.
CO6: To be comfortable to dine with colleagues, clients, and leaders comfortably in a formal or informal setting.

Unit 1: Accent Neutralization 6

Identifying and dealing with Mother Tongue Influence (MTI) – Pronunciation - Vowel Sounds and Consonant Sounds
– Inflection – Pausing - Reducing rate of speech - Volume and tone – Pitch – Clarity - and enunciation.

Unit 2: Customer Service 6

Customer Service - Different types of customers - Difference between customer service and customer experience -
Telephone Etiquette - Handling difficult customers.

Unit 3: Problem Solving and Decision Making 6

Define a Problem - Define Decision Making- Blocks in problem solving - Stereotyping and unconscious biases - The
process of Problem Solving and decision making - Problem Analysis- Decision Analysis - Potential Problem /
Opportunity Analysis - Creative Thinking - Problem Solving process - Implementation of the solution.

Unit 4: Business Email Etiquette and Chat 6

Emails Etiquette: Share format/ signature - Emails etiquette - dos and don’ts.

Unit 5: Basics of Finance 6

Accounting systems and how transactions are recorded - Financial statements: Profit & Loss account - balance sheet
- cash flow statement - Fixed assets - depreciation and the capitalization of software development expense - Working
capital and cash management - Using ratio analysis to assess corporate health and performance - Funding the
business: equity - debt and other aspects - Budgeting & Forecasting – capex – apex - Designing a flexible budget -
Capital expenditure appraisal and approval

21
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Security for Data Science (Professional

Title Code
Elective – III)
Prerequisite Credits | Total Hours 3-0-0 [3] | 45

Course Outcome:
Students will be able to refresh the Fundamentals of Cyber Security

CO1: Understand the Fundamentals of Cyber Security.

CO2: Implement Secure Data Handling Practices
CO3: Analyse Security Risks in Data Science Projects
CO4: Develop Threat Detection and Response Strategies.
CO5: Design Ethical and Privacy-Preserving Data Science Solutions.

Unit 1: Introduction to Cyber Security and Data Science 9

Overview of Cyber Security and Data Science - Definitions and Concepts - Intersection of Cyber Security and Data
Science - Cyber Threat Landscape - Types of Cyber Threats - Attack Vectors and Techniques - Impact of Cyber Attacks
on Data Science Processes - Foundations of Data Science - Data Collection and Sources - Data Storage and Management
- Data Processing and Analysis Techniques.

Unit 2: Foundations of Cyber Security 9

Principles of Cyber Security - Confidentiality, Integrity, and Availability (CIA) - Authentication and Authorization -
Encryption and Cryptography - Secure Data Handling - Data Classification and Sensitivity - Data Masking and
Anonymization - Secure Data Transfer and Sharing - Data Privacy and Compliance - Privacy Regulations (GDPR, HIPAA)
- Data Governance and Compliance Frameworks - Ethical Considerations in Data Science and Cyber Security.

Unit 3: Data Privacy and Protection 9

Data Privacy and Protection -Secure Data Sharing and Transfer - Secure File Transfer Protocols - Secure Data Exchange
Platforms - Securing Data Collection Systems - Best Practices for Secure Data Storage - Cloud Security and Data Privacy
- Secure Data Transfer and Backup Strategies - Data Retention Policies and Compliance.

Unit 4: Threat Detection and Incident Response 9

Threat Detection and Incident Response - Security Information and Event Management (SIEM) - Log Management and
Analysis - Real-time Threat Detection - Incident Response Frameworks - Preparation, Identification, Containment,
Eradication, Recovery - Forensic Analysis Techniques - Machine Learning for Cyber Security - Threat Prediction and
Classification - Behavioural Analysis and User Profiling.

Unit 5: Advanced Topics in Cyber Security for Data Science 9

Advanced Topics in Cyber Security for Data Science - Adversarial Machine Learning - Evasion Attacks - Defence
Mechanisms - Secure Machine Learning Models - Privacy-Preserving Machine Learning - Federated Learning - Ethical
and Legal Considerations - Bias and Fairness in Cyber Security - Ethical Hacking and Responsible Disclosure.

22
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Application Architecture & Deployment

Title Code
(Professional Elective – III)
Prerequisite Credits | Total Hours 3-0-0 [3] | 45

Course Outcome:
Students to understand how architect and AI Application deployment with important aspects to be taken care of.

CO1: Understand the differences between monolithic and microservices architecture and their respective
advantages and disadvantages in AI applications.
CO2: Understand the basics of Kubernetes and how it can be used to manage and deploy AI models in a production
environment.
CO3: Understand application programming interfaces (APIs) and their role in integrating AI models into larger
systems.
CO4: Understand MLOps and how it can be used to streamline the machine learning lifecycle, from data preparation
to model deployment and monitoring.

Unit 1: Monolithic vs Microservices 9

Introduction to Software Architecture and its types - What is Monolithic Architecture and its Importance -
Characteristics of Monolithic Architecture - Limitations of Monolithic Architecture - What are Microservices -
Working of Microservices - Main Components of Microservices Architecture - Advantages of Microservices -
Monolithic vs Microservices - Real World Example of Microservices - Challenges in Microservices.

Unit 2: Application Programming Interface 9

What is an API - How do an API Work - WEB APIs - LOCAL APIs - PROGRAM APIs - SOAP, REST API - What are REST
APIs - HTTP methods (GET, POST, PUT, DELETE) - Status Codes and URI structure - SOAP vs REST - What is API testing
- Types of Testing - Tools for API Testing - Authentication Mechanisms - Authorization Mechanisms - Role Based
Access Control (RBAC)

Unit 3: Containers - An Introduction 9

What is Virtualization - Virtualization in Cloud Computing - Introduction to containerization - Container Lifecycle -

Virtualization vs Containerization - Container Security - Serverless Containers - Introduction to Docker - Docker
Architecture - Components of Docker - Concept of Docker Images - Docker Commands - Advantages of Docker -
Introduction to Orchestration tools

Unit 4: Kubernetes - An Introduction 9

What is Kubernetes (K8s) - Why Kubernetes and not only docker - Kubernetes Components - Node - Control Plane -
Networking in Kubernetes - Kubernetes Resources - Pod, Deployment, Service, Volume, Namespace, node, cluster -
Storage - Security - Monitoring, Logging, Scaling - Writing YAML files.

Unit 5: ML Operations 9
Introduction to ML Operations - What is SDLC - Stages of SDLC - Waterfall Model - Agile Model - Iterative Model -
Importance of Each Models - Model Training - Model Deployment.

23
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Reading Materials:
1. Scott Surovich & Marc Boorshtein, Kubernetes and Docker – By Packt Publishing (2021)
2. Mark Treveil, Nicolas Omont & Clément Stenac, Introducing MLOps: How to Scale Machine Learning in the
Enterprise (Grayscale Indian Edition) – By Shroff/O'Reilly (2020).

24
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Gen AI Based Applications lab

Experiment No Experiment Details Type
Take any large language model (say GPT 3.5) and try to execute some query through
it. Create a small program where you can change the parameter values of
Experiment 1 Individual
Temperature, Top P and Max Tokens. Please identify how you can make your answer
more deterministic?
Please identify what are the basic metrices to evaluate your large language model
response? (As example, toxicity, biasness etc). Please write a short program where
Experiment 2 Individual
you can take model response as input and calculate the score for the above metrices
to understand output quality.
Please write a program where you can perform keyword-based search. Please take
Experiment 3 any text file as input and provide "keyword" dynamically and see whether your Individual
algorithm can search it effectively.
Please write a program where you take perform embedding based search. Please
take any vector database and use any embedding technique to search the answer of
Experiment 4 Individual
the query from the given input text file where query and text files are the inputs of
your program.
Please take 2/3 medical reports (may be blood reports) and store them in a place.
Experiment 5 Please write a program which can read all the files dynamically from the given Individual
locations. Please try to understand the metadata of the reports.
Create a set of questions for which you want to retrieve information from the
Experiment 6 medical reports through large language models. Save it in some database and keep Individual
in the excel file.
Apply large language model and Implement the RAG based approach to search the
Experiment 7 answer of the queries from the documents where two inputs will be taken: set of Individual
medical reports prepared in Experiment 5 and questions prepared in Experiment 6.
Perform the evaluation based on RAG-triad (Context Relevance, Groundedness and
Experiment 8 Answer Relevance). Show the importance of "context" towards getting the optimized Individual
output.
Use Palm 2 (or any other LLM) to perform automation of software development
Experiment 9 Individual
tasks which includes code generation, code debugging and test case generation.
Experiment 10 Use any diffusion model to generate images based on given prompt. Individual
Apply zero shot, one shot and few shot prompting and show how performance is
Experiment 11 Individual
improved in few shot prompting.
Apply chain-of-thought (CoT) in prompting and see how output accuracy increases.
Experiment 12 Do a comparison between normal prompting and CoT based prompting from output Individual
performance perspective.
Take a foundation model, create an instruction based fine tuning dataset, apply
Experiment 13 Individual
instruction fine tuning on the base model.
Perform performance evaluation of the model response between foundation model
Experiment 14 Individual
and after fine tuning it.
Experiment 15 Explore various task specific benchmark datasets and try to create a new one. Individual

25
©LTIMindtree | Privileged and Confidential 2024
 LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

Large Language Models Lab

Experiment No Experiment Details Type
Present you POV on the evolution of Large Language Models. Articulate their Group Project
Case Study 1
growth, architecture changes and application landscape
Present your POV on the different fine-tuning methodologies. Articulate the Group Project
Case Study 2
differences, the advantages, and disadvantages of each approach.
Case Study 3 Present your POV on the constitutional AI, how it’s different from RLHF. Group Project
Present your POV on the Quantization of LLMs, different techniques that are Group Project
Case Study 4 available, performance of the Quantized Models in comparison to the Original
Models
Present your POV on innovative architectures in transformer model that can lead to Group Project
savings in training or inference time. As an example, MoE from Mistral is one such
Case Study 5 unique architecture. Articulate tne performance of new architectures compared to
the original architectures and come up with some new architecture that can lead to
savings
Case Study 6 Present your POV on the Sustainable AI, Ethical AI, Trustworthy AI Group Project

26
©LTIMindtree | Privileged and Confidential 2024
LTIM –MTech AI-DS Syllabus
(Ver. 1.5 /12-06-2024)

27
©LTIMindtree | Privileged and Confidential 2024