CP Calendar

Built a customizable online calendar for competitive programming contests.

• Fetches contests from 120+ online judges using the Clist API
• Users can filter contests by platform and keywords to include or exclude events
• Generates a personal .ics calendar stream that syncs with Google Calendar or any app
• Created for the CP community after facing difficulty tracking contests across sites
• Hosted on PythonAnywhere and built using Flask, JSON, and frontend templating

This was my first public open-source project, inspired by a problem I personally faced.

BLANT

Worked on improving BLANT (Basic Local Alignment of Network Topology), a tool for analyzing complex networks through graphlet sampling, under Prof. Wayne Hayes at UC Irvine.

• Designed a recursive structure that reduced memory use for graphlet lookup tables from 180TB to under 4GB, enabling support for 10-node graphlets
• Developed a method to estimate absolute graphlet counts using star-motif-based approximations from relative concentrations
• Derived normalization formulas for NBE (Node-Based Expansion) and EBE (Edge-Based Expansion) sampling methods, making them usable in practice with reduced statistical noise
• Created a new sampling algorithm, SEC (Sequential Edge Chaining), combining low noise and high speed — nearly matching MCMC while offering independent samples
• Improved the α-computation algorithm in MCMC sampling, reducing runtime from days to seconds for k=8

ssehC

Created a chess engine from scratch in C as a personal challenge to understand how chess algorithms work.

• Based on principles from the VICE chess engine tutorial by BlueFeverSoft
• Implemented board evaluation, move generation, and minimax search with alpha-beta pruning
• Gained hands-on experience with low-level programming and debugging complex logic
• Deployed the engine as a Lichess bot, where it achieved a 2000+ rating

This project helped me appreciate both the beauty of chess and the complexity of building something that plays it well.

Spotify Addon

Improved an existing Java-based Spotify lyrics widget by expanding support for non-English music.

• Added a new synced lyrics API (lrclib.net) with a broader multilingual database
• Learned Java from scratch to understand and modify the open-source codebase
• Integrated and formatted API output for compatibility with the app’s UI and timing engine
• Contributed changes via pull request, which was accepted and merged by the original author

The project now supports Bengali and other regional songs that were previously unsupported.