Building friendly frontends, reliable backends and ship end-to-end features, effectively and efficiently.
Projects
Selected work across ML and full‑stack projects.
Path Planning & Control Sandbox
Path planning is abstract without visuals → Implemented interactive A* planning with Chaikin smoothing and Pure Pursuit/PID tracking.
Toronto Safety Dashboard
Created a user-friendly dashboard to explore Toronto Police Service MCI near any address — postal‑code search, time and radius filters, and colour legend.
LSTM Classical Music Generator
Symbolic‑music LSTM that generates multi‑bar classical‑style phrases; end‑to‑end MIDI pipeline from parsing → training → synthesis.
Encrypted File Sharing
Local storage lacked security → built an encrypted file system in Go (AES-GCM, HMAC) for safe sharing.
F1 ｘ ML Project (WIP)
🚧 Under construction 🚧 🏎️💨 Experimenting with racing data + ML predictions. Demo + repo coming soon.
k‑NN Geolocation
Image geolocation via CLIP embeddings + k‑NN; grid‑searched k and PCA analysis delivered the lowest MDE in my experiments.
cvmoji - emoji resume generator
Plain text resumes are boring → created a React/TypeScript generator that maps skills to emojis and exports SVG/PNG.
LinkedOut
LinkedIn titles too plain? Type your role → get a cursed lineup of inflated, buzzword-heavy titles (with Meme Mode for extra LinkedIn clout).
Shift Scheduler
Constraint‑aware matching automates data collection and shift assignments, improving scheduling accuracy and cut scheduling time by 30%.
Experience
- Filing errors led to high customer support volume. Built and deployed Spring Boot REST API endpoints, improving data accuracy and reducing support calls by over 15%.
- Data inconsistencies between backend and UI disrupted workflows. Integrated APIs into customer-facing UI and refined schemas, ensuring reliable transactions across thousands of daily filings.
- Distributed systems required tamper-proof validation. Implemented Merkle tree data structures in Go/Rust with MongoDB integration, reinforcing backend security.
- Automated operational task workflows with secure scripts, cutting maintenance time by ~30%.
- Supported 50+ students mastering core data structures and algorithms via 1:1 guidance and debugging, improving lab completion rates by ~25%.
- Manual shift scheduling for 80+ staff was slow and error-prone.
- Took initiative to develop a shift scheduling tool, reducing scheduling time by >30% and boosting department efficiency.