• Developed a comprehensive financial analysis platform using Python, integrating traditional statistical methods with advanced machine learning algorithms to analyze stock market data, implementing scikit-learn regression models for correlation analysis and statsmodels for cointegration testing across multiple asset classes
• Engineered automated data pipeline and visualization system leveraging Yahoo Finance APIs, pandas for data manipulation, and matplotlib/seaborn for generating professional-grade financial charts including correlation heatmaps, volatility analysis, and regression scatter plots with statistical significance testing
• Built modular ML architecture featuring pairwise regression analysis with beta coefficient calculation, rolling correlation matrices, and Engle-Granger cointegration tests, resulting in actionable investment insights through R² modeling and risk-adjusted performance metrics (Sharpe ratio, skewness, kurtosis)

CodeClash lets you challenge friends or AI bots to real-time 1v1 coding duels, where the first to solve the problem wins. You can track your progress, climb ranks from Bronze all the way to Grandmaster, and see how you compare on global leaderboards. Beyond just duels, users can customize their profiles, choose avatars, display skills, and set goals — making practice more personal and engaging. It’s a way to practice coding, compete with friends, and have fun doing it.

My contributions:

• Architected real-time multiplayer system using Node.js, Express, and Socket.io to enable synchronized 1v1 coding duels between users, ensuring consistent game state and preventing desyncs across concurrent WebSocket connections
• Implemented secure user authentication and authorization with JWT tokens supporting multiple login methods (Google OAuth, email), while maintaining session security and seamless user experience across the platform
• Integrated Gemini API to power intelligent bot opponents for solo practice matches, creating adaptive AI competitors that simulate realistic coding challenge scenarios and provide users with engaging single-player gameplay options

• Collaborated in a team of two to analyze a dataset, focusing on NYC vehicle accidents from 2016 to 2022
• Conducted thorough data exploration and cleaning to ensure high-quality data for analysis and modeling
• Trained and visualized the model using multiple python libraries and achieved a RMSE of ~ 0.38%
• Presented findings and model results to the class, demonstrating a deep understanding of the dataset and the analytical techniques employed
• Demonstrated strong teamwork and communication skills, coordinating effectively to meet project deadlines and deliver high-quality results

This project provides a simple yet efficient video streaming server for the Raspberry Pi, enabling users to upload and stream videos through a web interface. Its core features include user authentication, secure video storage, a RESTful API for managing media, and a built-in video playback interface. The application relies on Python 3.8+, PostgreSQL for data storage, and FFmpeg for video processing. A Raspberry Pi Model 3B+ or later is recommended, and full setup instructions can be found in the repo.