Software Engineer
MSCS @ Columbia. Software engineer working at the intersection of mobile, AI, and user experience with a creative background in music.
A privacy-first, fully local AI conversational therapy companion designed to run entirely on-device with no cloud APIs. The app combines real-time speech recognition, a locally hosted large language model, and natural text-to-speech to deliver supportive, voice-based conversations while ensuring that no user data ever leaves the machine. Built for Snapdragon PCs, it demonstrates how an end-to-end AI pipeline β ASR, LLM, and TTS β can be optimized for consumer hardware to provide accessible, stigma-free mental health support with strong privacy guarantees.
iOS app for vocal training that includes customizable vocal exercises and vocal warmups with full control over pitch range and speed. It features high-quality piano sounds -- offering a professional accomponiment experience from your phone.
iOS app focused on tracking added sugar and making sure sugar level is not too high.
Climbing Training tracking and analytics iOS app for boulderers and sport climbers.
Interactive mobile musical instrument with touch controls and gyroscopic pitch bending. A web-based tool that transforms your phone into a creative sound device.
RAG-based AI chatbot trained on voice textbooks, providing specialized vocal training advice and techniques for singers and voice coaches.
A Streamlit web app that fetches and exports metadata from Spotify tracks, albums, artists, and playlists. Built to work with Spotify's Development Mode restrictions (post-November 2024).
Hardware synthesizer with 9 buttons creating unique sounds through different button combinations. Built with Arduino and custom C++ programming.
3D interactive musical instrument built with Three.js and Tone.js. Drag spheres in 3D space to create music and explore spatial audio relationships.
LightGBM-based TSLA forecasting with engineered features, market context (SPY, VIX), quantile intervals, and a daily runner for automated nextβday and 5βday predictions. Includes the original research MVP plus a modular pipeline with persistence, logging, and CLI runners.
I'm always excited to collaborate on innovative projects in mobile and web.