2025
Sherlock
High-throughput job data aggregation platform ingesting 15K+ records/day from 160+ sources with 20+ concurrent workers, content-hash deduplication, and Redis-backed caching.
- Python
- Redis
- Selenium
- GraphQL
More stuff I've built. Some for fun, some out of necessity, all functional. Plenty more sitting in private repos.
High-throughput job data aggregation platform ingesting 15K+ records/day from 160+ sources with 20+ concurrent workers, content-hash deduplication, and Redis-backed caching.
Real-time waste classification with YOLOv8 + ONNX Runtime achieving 90% accuracy at 33ms latency (30 FPS on CPU).
Backend workflow automation compiling natural-language requirements into executable pipelines across 30+ node types via 15+ REST endpoints.
Firefox extension persisting DevTools inspect-element edits across reloads with local-only storage.