A full-fledged product (frontend dashboard, backend pipelines) that sources job postings daily and scores each one against real experience with evidence, not keywords. Built to minimize LLM token usage and plug into whichever model you already use, Claude or ChatGPT. The goal: take the boring part of a job search off your plate.

Self-built, self-hosted, already running daily for a few friends' searches.

Built Sama's Sensor Fusion platform: $7M in revenue impact, bookings share up from 3% to 45%. Also built AI-assisted tools (a 3x efficiency gain).

10+ international keynotes and panels on LiDAR annotation quality, ADAS/AV data labeling, and training data safety, at Auto AI USA, Tech AD Europe, AVT Expo, and others.

Owned API connectivity for data ingestion: native, purpose-built connectors alongside push-API and generic REST-API gates for systems without one, covering both sides of the native-vs-generic, push-vs-pull tradeoff.

Before being hired, scraped every public Breather coworking location and built a full analysis flagging underperforming locations to double-check. The team hired me off the back of it, opening a new data role for marketing and sales, even though the position I'd originally applied for was already filled.

Full rebuild of the company website (+50% sessions, −30% bounce rate) and a new online quotation widget (+50% conversions, +30% sales). The site has stood as the business's main lead-generation channel since launch. Still true today.