Computer vision and ML engineer. I build production systems end to end: data collection, labeling, training, evaluation, deployment.
Currently: sole engineer at Pier 60 (Clearwater Beach, FL — Tripadvisor's #2-ranked U.S. beach). I pitched the CEO on automating rental occupancy tracking and was hired on the spot. The system I built and ship weekly covers ~1,000 cabanas and umbrellas, runs occupancy checks every 2 minutes instead of the old 2-hour walking sweeps, and drives a 17%+ lift in daily rental revenue.
Demo video · Portfolio · anthonyhayward1000@gmail.com
M.I.L.E.S — One of the first agentic LLM voice assistants on GitHub (2024). 232 stars, 29 forks. Tool orchestration, persistent memory, system and Spotify control, plus a custom wake-word model trained on 50,000 samples. All inference runs locally.
memorium — Local-first screen memory. Continuous capture, OCR, embeddings stored in SQLite, semantic search with voice recall. Fully on-device. Python, React, Tauri.
MacroScan — iOS nutrition app, shipped to the App Store solo. Photo-to-macros estimation with multi-provider AI routing, scan history, and subscriptions. React Native, Expo.
crypto-one — Transformer trading research over 15-minute SOL market data. The part worth reading: the evaluation harness caught data leakage that invalidated my own best backtest results before live deployment. The forensics are documented in the repo.
Smaller: objectTrack (OpenCV tracking and re-identification) · handTrack (gesture mouse control) · attendify (club attendance platform) · Fuel-Up (iOS fuel discovery)
Python, TypeScript, PyTorch, OpenCV, MediaPipe, Sentence Transformers, React, React Native (Expo), Node.js, Tauri, SQL, Firebase, Supabase
Open to computer vision / ML engineering roles — anthonyhayward1000@gmail.com