Technical Projects

1. Reinforcement Learning in Billiards, Football, and Cricket

CS 747: Foundations of Intelligent and Learning Agents, Prof. Shivaram Kalyanakrishnan

Designed agents utilizing Monte-Carlo Tree Search (MCTS) in Billiards, developed optimal strategies through MDP-Planning for scoring goals in Half-Field Offense Football, and executing a successful run chase in cricket. Read More


2. Deep Recurrent Q-Learning for Partially Observable MDPs

EE 782: Advanced Topics in Machine Learning, Prof. Amit Sethi

This project presents a unique implementation of Deep Recurrent Q-Network (DRQN) that incorporates Transfer Learning for feature extraction, a customized LSTM for temporal recurrence, and a domain-informed reward function for flickering Atari 2600 games. Read More


3. Online Learning

CS 747: Foundations of Intelligent and Learning Agents, Prof. Shivaram Kalyanakrishnan

Derived and implemented an asymptotically optimal algorithm based on Bayesian inference for a faulty Multi-Arm Bandit instance. Implemented variants of Thompson Sampling and KL-UCB for solving a Batched Multi-Arm Bandit Problem. Read More


4. Rapid Real-Time Multi-Face Detection

EE 769: Introduction to Machine Learning, Prof. Amit Sethi

As a part of the endterm project for the course, developed a rapid real-time multi-face detector that utilizes “Integral Image Representation” for rapid computation of Haar features, subsequently used by Adaboost for classifier learning. Read More


5. Generative AI & LSTM-Based Stock Trading System

EE 782: Advanced Topics in Machine Learning, Prof. Amit Sethi

Implemented a Siamese Model to enhance a CGAN for generating diverse images of a given individual, and I also developed an LSTM-based stock trading system for high-frequency trading, refining it to incorporate multiple stock prices. Read More


6. Implementation of Control Systems

EE 324: Control Systems Laboratory, Prof. Dwaipayan Mukherjee

Implemented a PID controller for a line follower Spark V robot, designed active noise-cancellation circuit for a pair of headphones and achieved precise DC motor position control using embedded feedback controller. Read More


7. Biomedical Image Segmentation Using Deep Learning

EE 610: Image Processing, Prof. Amit Sethi

Utilization of a multi-organ transfer learning approach using U-Net architecture tailored for binary semantic segmentation for distinguishing between nucleus and non-nucleus regions in stained tissue images. Read More


8. Multi-Faceted Image Restoration and Enhancement

EE 610: Image Processing, Prof. Amit Sethi

Wide-ranging and challenging problem statements which dealt with implementing a research-based technique properties of neighbouring wavelet coefficients for image denoising, image restoration of motion-blurred cars to effectively reveal license plate characters, and satellite image enhancement. Read More


9. Raptor Codes

EE 605: Error Correcting Codes, Prof. Nikhil Karamchandani

Simulated digital message transmission utilizing Raptor codes and delivered a concise talk showcasing the significance of Raptor codes as a contemporary coding technique as a component of the course endterm project. Read More


10. Algorithmic Design of Communication Systems

EE 341: Communication Systems Lab, Prof. Jayakrishnan Nair

Evaluated the performance of Costas Loop and Viterbi-Viterbi algorithms in discrete-time for phase and frequency offset removal in incoming signals. Implemented equalization filter design to negate the effect of ISI (Intersymbol Interference) on the digitally transmitted message suffering multipath reflection with the help of GNU Radio. Read More


11. Six-Stage Pipelined Microprocessor

EE309, Microprocessors, Prof. Virendra Singh

Designed the architecture of a RISC microprocessor which included all the hazard mitigation techniques like forwarding and branch prediction. The architecture allowed predicated instruction execution and multiple load and store execution. Read More