Projects
non-comprehensive, still working on adding everything...
MarketMentor
Hacktech'25 Finalist
AI-powered financial news clustering system using OpenAI embeddings and Flask APIs to generate market predictions with confidence scores and personalized portfolio insights. Implemented real-time analysis pipeline processing AlphaVantage/Polygon data with sub-second response times. Built React/TypeScript dashboards visualizing event relationships and risk/opportunity signals.
Humanizing the MTA
Cubist'24 Finalist
Interactive transit safety platform with real-time data visualization. Built with Flask and Folium for geospatial mapping, uses scikit-learn autoregression models to predict subway crowding and safety metrics from NYC crime and MTA APIs.
Reddit Real-Time Sentiment Analysis
NLP Full-Stack Web Application
Real-time sentiment analysis platform for Reddit subreddits using NLP. Computes average and median sentiment of subreddit posts. Uses Reddit API (PRAW) and fine-tuned DistilBERT model hosted on Hugging Face Inference Endpoints, cached with Redis.
Mini Trading System
High-Performance Order Book Engine
Limit order book exchange sim w/ optimal DSA for low-latency trading. Uses hybrid indexing: std::map (red-black tree) for O(log L) price lookups with O(1) best-bid/ask, std::list for O(1) FIFO operations with stable iterators, and unordered_map for O(1) cancel-by-ID. Read the /planning_docs for thought processes.
Amazon E-Commerce Simulation
CS316 Databases Final Project
Full-stack e-commerce platform with secure authentication, catalog management, seller inventory, and transaction processing. Optimized PostgreSQL schema with strategic indexing reduced query response times from 800ms+ to 150ms. Built RESTful API architecture in Flask handling concurrent financial transactions with distributed processing capabilities.
