Projects

A client-side stock-analysis tool: enter a US ticker to get support/resistance zones, entry/stop/target levels, risk:reward, and position sizing. All math runs in the browser and shows its work — the formula and source behind every number. Not financial advice.

An AI agent using Minimax with alpha-beta pruning to play checkers. Built for the CSCI-561 "Foundations of AI" course, where it competed against other students' agents.

Classification, regression, and clustering algorithms — plus metrics, preprocessing, and model-selection helpers — implemented from scratch with NumPy for a deeper understanding of how they work.

A U-Net (from "U-Net: Convolutional Networks for Biomedical Image Segmentation") built in Keras to segment brain tumors in MRI scans.

An LSTM-based seq2seq model that tags every word of a paragraph with its Named Entity or Part of Speech. Served with Flask and Docker.

The "merge" architecture from "What is the Role of RNNs in an Image Caption Generator?" implemented in Keras and deployed with gRPC and tf-serving on Docker.

A Kaggle challenge: given a pair of face images, decide whether the two people are related. Tackled with a Siamese network over VGG-Face.

Inspired by thispersondoesnotexist.com. A Deep Convolutional GAN trained on 100k celebrity photos whose generator produces novel human faces.

A reinforcement-learning agent trained with Deep Q-Learning on a Double Dueling network with Prioritized Experience Replay to play Flappy Bird. Implemented in PyTorch.

A Siamese network in Keras for one-shot face recognition — recognizing faces without the extensive training samples a classifier would need.

A Kaggle image-segmentation competition: identify pixels containing salt sediment in seismic images. Solved with a U-Net encoder–decoder.

An attention-based machine-translation model that converts human-readable dates to a machine-readable format. Inspired by the final week of Andrew Ng's Deep Learning Specialization.

A web app where you draw a digit on an HTML canvas and a deep ConvNet predicts what you drew. Hosted on Flask.

A chatbot that returns the meaning of any word via the Oxford Dictionary API. Built with Amazon Lex (intents, slots, prompts), AWS Lambda, and Slack.

The classic 8-puzzle playable in a Linux terminal. Built with python-curses and solved with A* search using Manhattan distance as the heuristic.

A terminal Snake game written in Python 3 in about 200 lines — comments, whitespace, and credits included.

Built at Smart India Hackathon 2017 (2nd place nationally, Dept. of Biotechnology). An Android app that authenticates exam candidates via Aadhaar biometrics to prevent impersonation.

An Android app that tracks friends via GPS and notifies you when one is nearby, using the Google Maps Distance Matrix API. Tracking is opt-in per invitation.