Resume

• ▪AI-based medical image analysis under Prof. Danny Chen

• ▪Owned two client products end to end as sole engineer, from scoping in weekly client calls through architecture and deployment
• ▪Built and shipped DEEM's EPA-compliance refrigerant-management web app in Next.js and Fastify over SQL Server, delivering a certification-gated purchase-order system with automated inventory drawdown and reconciliation plus mobile barcode scanning, tracking 40,000+ lbs of refrigerant a year against $59K-per-day EPA penalties
• ▪Drove test-driven development with a 200+ test integration suite on a Dockerized SQL Server, and built GitHub Actions and Vercel CI/CD to auto-deploy on merge
• ▪Engineered a full-stack job-scheduling and profitability platform in Next.js, TypeScript, and Supabase for a landscaping business launching August 2026, replacing quote-by-feel estimating with per-job margin analytics aimed at lifting the owners' margins from 15-20% toward their 40% goal, and automating crew scheduling and SMS alerts via cron and Twilio

• ▪Re-engineered the protected health information (PHI) browser used to manage patient records, migrating the full UI from Java Swing to JavaFX and aligning it with the main application's design system
• ▪Built encryption, anonymization, and USB-export features for patient records, hardening handling of sensitive medical data to support HIPAA-compliant workflows
• ▪Integrated PACS (Picture Archiving and Communication System) server functions into the browser's export pipeline, partnering with the Head of IT to connect clinical imaging infrastructure

Languages

Frameworks & Libraries

Tools & Platforms

• ▪Built a full-stack platform (FastAPI, PostgreSQL, Next.js) that combines personal Strava, WHOOP, and Dexcom G7 glucose data to generate personalized training plans and coaching using the Anthropic API
• ▪Engineered an async FastAPI backend integrating four external REST APIs via OAuth 2.0; implemented JWT authentication with bcrypt-hashed passwords and per-user data isolation

• ▪Developed a PyTorch stock-price prediction model combining historical market data (yfinance) with FinBERT news-sentiment features engineered from large financial and news datasets using pandas

• ▪Shipped an upstream-merged safety feature adding a collapsible diff preview (diffstat + unified diff) to revert confirmation dialogs, preventing users from irreversibly discarding uncommitted work
• ▪Shared path selection between preview and checkout so the two cannot diverge; extended MessageBox with an expand_details API and added unit tests for the new gitcmds diff helper

• ▪Led a team of six junior counselors supporting campers aged 7–15 in managing type 1 diabetes, mentoring daily self-care while maintaining camp operations across multiple summer sessions

• ▪Self-taught DJ performing campus gigs; built custom production speakers from scratch with crossovers, woofers, and cabinets

• ▪Completed Technical Writing and Professional Ethics coursework; proposed software improving Rome's bus-arrival predictions