M.Tech AI/ML Academic Portfolio

BITS Pilani M.Tech Program - Comprehensive AI/ML Journey

A comprehensive collection of AI/ML projects, assignments, and research implementations from my M.Tech program at BITS Pilani. This repository demonstrates practical applications across Deep Learning, Reinforcement Learning, Natural Language Processing, MLOps, Computer Vision, and Generative AI.

---

Table of Contents

• Overview
• Repository Structure
• Projects Showcase
  • Semester 3: Advanced Systems, Computer Vision & MLOps
  • Semester 2: Deep Learning & NLP
  • Semester 1: Foundations & Game AI
• Technologies Used
• Academic Progress
• Setup & Usage
• Contact

---

Overview

This repository showcases my academic journey through advanced AI/ML concepts, transitioning from foundational algorithms to production-ready MLOps pipelines and Generative AI systems.

Key Highlights:

• End-to-End MLOps: Full lifecycle management using Docker, Kubernetes, and CI/CD pipelines.
• Computer Vision: Classical feature engineering and cross-modal attention for semantic segmentation.
• Generative AI: Implementation of RAG systems and LLM integration.
• Advanced NLP: From statistical translation to modern transformer-based architectures.
• Reinforcement Learning: Solving complex environments with Actor-Critic and DQN agents.

---

Repository Structure

academic-ai-ml-portfolio/
├── semester-3/                       # Advanced Systems, GenAI & MLOps
│   ├── computer-vision/             # Image Classification & Segmentation
│   ├── conversational-ai/           # Hybrid RAG & LLM Applications
│   ├── mlops/                       # Production ML Pipelines (Docker/K8s)
│   └── nlp-applications/            # SMT & Web-based NLP Tools
├── semester-2/                       # Deep Learning & Core NLP
│   ├── deep-neural-networks/        # CNN/DNN Architectures
│   ├── deep-reinforcement-learning/ # RL Algorithms (DQN, PPO)
│   └── natural-language-processing/ # Financial Sentiment Analysis
├── semester-1/                       # Foundations
│   └── game-ai-minimax/             # Strategic Game AI
└── environments/                     # Configuration & Setup

---

Projects Showcase

Semester 3: Advanced Systems, Computer Vision & MLOps

1. Binary Image Classification – MLOps Pipeline (Cats vs Dogs)

Domain: MLOps | Location: semester-3/mlops/assignment2

• Description: An end-to-end MLOps pipeline covering data versioning (DVC), experiment tracking (MLflow), FastAPI inference, containerization (Docker), and CI/CD via GitHub Actions.
• Tech Stack: TensorFlow/Keras, MLflow, DVC, FastAPI, Docker, GitHub Actions, Prometheus.

2. Computer Vision: Image Features & Classical Machine Learning

Domain: Computer Vision | Location: semester-3/computer-vision/assignment-1

• Description: Explores foundational computer vision techniques for satellite image classification. Emphasizes classical handcrafted feature engineering (LBP, HOG, Edge Detection) paired with robust ML models (SVM, Random Forests).
• Tech Stack: OpenCV, scikit-learn, Scikit-Image, Feature Engineering

3. Computer Vision: Cross-Modal Attention & Semantic Segmentation

Domain: Deep Computer Vision | Location: semester-3/computer-vision/assignment2

• Description: Implements advanced deep learning architectures for semantic segmentation and complex object detection. Centers on cross-modal attention frameworks for predictive environmental monitoring (Sentinel-5P NO2 dataset).
• Tech Stack: PyTorch, Faster R-CNN, Semantic Segmentation, Attention Mechanisms

4. Hybrid RAG Question Answering System

Domain: Generative AI | Location: https://github.com/anandsuraj/hybrid-rag-system-with-automated-evaluation

• Description: A sophisticated RAG (Retrieval-Augmented Generation) system that combines dense (FAISS) and sparse (BM25) retrieval to answer questions from a dynamic Wikipedia corpus. Features Reciprocal Rank Fusion (RRF) for optimal context retrieval.
• Tech Stack:
  • LLM: Flan-T5-base (Instruction Tuned)
  • Vector DB: FAISS (Dense Retrieval)
  • Search: Rank-BM25 (Sparse Retrieval)
  • Backend: Flask

5. Heart Disease Prediction MLOps Pipeline

Domain: MLOps & Healthcare | Location: semester-3/mlops/assignment1/mlops-heart-disease

• Description: A production-ready Machine Learning pipeline for heart disease prediction, demonstrating end-to-end MLOps practices. Includes containerization, orchestration, and automated testing.
• Tech Stack:
  • Infrastructure: Docker, Kubernetes (K8s)
  • ML Framework: Scikit-learn
  • Tools: PyTest, CI/CD workflows

6. Statistical Machine Translation (SMT) System

Domain: NLP Applications | Location: https://github.com/anandsuraj/nlp-statistical_machine_translation_with_bLEU_evaluation

• Description: A web-based translation workbench supporting multiple languages with integrated BLEU score evaluation. Allows comparison of translations against reference texts with detailed n-gram precision analysis.
• Tech Stack:
  • Core: Python, Flask
  • Evaluation: BLEU (SacreBLEU), N-gram analysis
  • API: Google Translate API

7. Intelligent Spell Checker

Domain: NLP Applications | Location: semester-3/nlp-applications/spell-checker-web-app-flask

• Description: A Flask-based web application for real-time spell checking. Features a history tracking system to learn from past errors and provides context-aware suggestions.
• Tech Stack: Flask, PySpellChecker, JSON Storage

---

Semester 2: Deep Learning & NLP

6. Financial Sentiment Analysis Engine

Domain: NLP | Location: semester-2/natural-language-processing/assignment1

• Description: An NLP system designed to analyze financial text data. It compares Skip-gram and CBOW word embedding models to classify market sentiment, aiding in automated trading decisions.
• Tech Stack: NLTK, Gensim (Word2Vec), Scikit-learn, Pandas

7. Sepsis Treatment Optimization & Drone Battery Management

Domain: Deep Reinforcement Learning | Location: semester-2/deep-reinforcement-learning/assignment1

• Description: Application of Actor-Critic RL algorithms to optimize sepsis treatment strategies in ICU settings, alongside a Drone Battery Management system using DQN/DDQN for autonomous surveillance.
• Tech Stack: PyTorch, OpenAI Gym, Stable-Baselines3

8. Deep Neural Networks: Architecture Design & Optimization

Domain: Deep Learning | Location: semester-2/deep-neural-networks/assignment1

• Description: A comprehensive study of Neural Network architectures on the MNIST dataset. Implementation and comparative analysis of regularization techniques (Dropout, L2) and depth variations to optimize performance.
• Tech Stack: TensorFlow/Keras, NumPy, Matplotlib

---

Semester 1: Foundations & Game AI

9. Strategic Game AI: Crossword Puzzle with Minimax Algorithm

Domain: Game Theory | Location: semester-1/game-ai-minimax

• Description: An AI agent capable of playing a strategic two-player crossword game. Implements the Minimax algorithm with depth-limited search to make optimal moves against human or AI opponents.
• Tech Stack: Python, Search Algorithms, Game Theory

---

Technologies Used

Domain	Technologies
Generative AI	Transformers (Hugging Face), RAG, FAISS, LangChain concepts
MLOps	MLflow, DVC, Docker, Kubernetes, CI/CD, PyTest, Prometheus
Computer Vision	OpenCV, Scikit-Image, Faster R-CNN, Segmentation Masks
Deep Learning	TensorFlow, Keras, PyTorch
NLP	NLTK, Spacy, Gensim, BLEU, Word2Vec
Reinforcement Learning	OpenAI Gym, Stable-Baselines3, Ray RLlib
Backend/Web	FastAPI, Flask, Python, JSON
Data Science	Pandas, NumPy, Scikit-learn, Matplotlib

---

Academic Progress

• Semester 1: Foundations (Algorithms, Python, Game AI)
• Semester 2: Core AI (Deep Learning, RL, NLP)
• Semester 3: Advanced Systems (Generative AI, MLOps, Computer Vision, Applied NLP)
• Semester 4: Dissertation & Research (Upcoming)

---

Setup & Usage

Prerequisites

# Python 3.8+
python --version

# Docker (for MLOps projects)
docker --version

Installation

# Clone repository
git clone https://github.com/anandsuraj/mtech-ai-ml-journey.git
cd mtech-ai-ml-journey

# Create virtual environment
python -m venv venv
source venv/bin/activate

Running a Project (Example: Hybrid RAG)

cd semester-3/conversational-ai/assignment/assignment-2
pip install -r requirements.txt
./run.sh

---

Contact

Suraj Anand
M.Tech AI/ML, BITS Pilani

surya13493@gmail.com
LinkedIn | GitHub

---

Created as part of the M.Tech AI/ML academic curriculum.