Projects

9 projects found

CVEContext

Aug 2026 – Sep 2026

.NET 10 Blazor app on Azure App Service that scans npm dependencies for known vulnerabilities.

  • Built a .NET 10 Blazor app on Azure App Service to scan npm dependencies for known vulnerabilities.
  • Integrated the GitHub REST API and OSV.dev to match dependencies against CVEs by version.
  • Correlated vulnerable-package imports to source files, using Azure OpenAI for prioritized risk tiers.
View on GitHub →

InterviewIQ (AI Hackathon 2026)

Jun 2026

Real-time AI interview simulator that grades users on their hiring decisions, built at UC Berkeley's AI Hackathon.

  • Built a real-time AI interview simulator (Next.js, TypeScript) that grades users on hiring decisions.
  • Architected a multi-agent Claude workflow that generates, role-plays, and grades realistic candidates.
  • Engineered a live Deepgram voice pipeline (Nova-3 STT, Aura-2 TTS) over WebSockets with barge-in.
View on GitHub →

Huddle (nwHacks 2026)

Jan 2026

Private family-oriented social platform with real-time interactivity.

  • Built full-stack application using Next.js (TypeScript) and Supabase with PostgreSQL.
  • Implemented authentication, media posting, and real-time presence tracking.
  • Developed a daily memory game with persistent leaderboard functionality.
  • Designed an accessible, mobile-friendly UI to promote engagement.
View on GitHub →

CodeBuddy

Oct 2025 – Jan 2026

Full-stack empirical performance analysis platform for JavaScript programs.

  • Built React + Node.js/Express app on AWS EC2 with nginx, pm2, and Vercel rewrites for HTTPS proxy.
  • Sandboxed execution in an isolated VM with per-IP rate limiting (30 req/min) for untrusted code.
  • Profiled runtime and memory, fitting results to a custom regression engine across 6 complexity classes.
  • Created interactive Chart.js visualizations to analyze performance trends across scaled inputs.
View on GitHub →

Must Dance (HackCamp 2025)

Nov 2025

Machine learning–powered dance scoring web app built in a 24-hour hackathon.

  • Co-developed React frontend and Flask backend to process and evaluate user-recorded dance videos.
  • Built a custom pose-based scoring system on top of MediaPipe landmarks to evaluate movement accuracy.
  • Integrated FFmpeg and OpenCV for video processing, frame extraction, and preprocessing pipeline.
  • Shipped a working MVP within 24 hours in a 3-person team.
View on GitHub →

Java Gym Progress Tracker

Jul 2025

Object-oriented workout tracking application built with MVC architecture.

  • Designed domain models including Workout, Exercise, and WorkoutLog with clean separation of concerns.
  • Implemented JSON-based persistence for saving and loading structured workout data.
  • Developed console-based interface followed by Java Swing GUI without modifying core model layer.
  • Integrated one-rep max prediction logic using strength estimation formulas.
View on GitHub →

Pac-Man Game

Jun 2025 – Sep 2025

Object-oriented Python recreation of the classic arcade game.

  • Implemented grid-based movement and custom collision detection engine.
  • Built enemy AI using BFS graph traversal for pathfinding.
  • Structured game entities using encapsulated OOP design principles.
  • Designed modular game loop with state management.
View on GitHub →

2048 Game

May 2025

Matrix-driven 2048 clone exploring algorithmic game logic design.

  • Implemented tile merging and shifting logic using 2D array transformations.
  • Separated core game logic from Pygame animation layer.
  • Built clean, modular object-oriented architecture.
  • Focused on deterministic state transitions and testing edge cases.
View on GitHub →

Predicting Newsletter Subscriptions with KNN

Apr 2025

Supervised machine learning project evaluating predictive power of user behavior features.

  • Built KNN classifier using tidymodels in R.
  • Performed data cleaning, upsampling, and 5-fold cross-validation.
  • Evaluated accuracy, precision, and recall metrics.
  • Analyzed weak predictive signal across selected behavioral features.
View on GitHub →