📍 Amsterdam · 🤖 AI Engineering @ Booking.com · 🦞 Building ClawMetry

I build AI/ML systems that ship. From LLM evaluation frameworks running across thousands of agents in production, to open-source observability tools for autonomous agents. I like first-principles products with no bloat, and I care about shipping over polishing forever.

• 🦞 ClawMetry (open source) : Real-time observability dashboard for AI agents. pip install clawmetry && clawmetry, zero config, read-only, DuckDB-first. Tracks cost, tokens, and behavior across 12+ agent runtimes (OpenClaw, NVIDIA NemoClaw, Claude Code, Codex, Cursor, Aider, Goose, and more).
• 🔁 flywheel.md (open standard) : The loop your AI coding agents follow to ship a change, prove it in production, and improve. A companion file to AGENTS.md and SOUL.md. (repo)
• 🖥️ ClawMetry for Mac (open source) : Native macOS menu-bar app for ClawMetry. Your agent fleet's cost and health, one click away in the menu bar.
• ⚖️ Judge LLM Evaluation Framework : Automated LLM-as-judge evaluation that replaced manual QA at scale. Powers evaluation across 15,000+ agents and delivered roughly 7x faster iteration cycles.
• 🕉️ VedicVoice : A Sanskrit digital library with AI-powered text-to-speech, giving voice to classical texts.
• 🎧 HeyMaya : A voice-controlled AI music DJ with a natural-language interface, built end to end in a week.
• 🚗 Neural-Net Self-Driving RC Car : A Raspberry Pi car that learned to drive itself with a neural network, back in 2012.

• Scaling LLM evaluation in production at Booking.com, driving customer-service AI accuracy from around 60% to 95%+ with automated eval loops instead of manual review.
• ClawMetry, making AI-agent cost and behavior observable in real time. The meter for the prompt-cache re-read tax and per-runtime spend across a whole fleet.
• Agent infrastructure : enforcement proxies, budget guardrails, and end-to-end encrypted cloud sync for teams running many agents.
• Side quests at the intersection of AI and culture, including Sanskrit TTS, voice interfaces, and hardware companions.

• AI Engineering / Tech Lead, Booking.com (2020 to present). AI/ML for Customer Service, LLM evaluation at scale.
• Lead Engineer, ClassKlap / Eupheus Learning. EdTech platform serving 500,000+ students.
• Early Engineer (#3), Recruiterbox / JazzHR. Recruiting SaaS, ground-floor engineering.
• B.E. Computer Science, Dr. Ambedkar Institute of Technology.

If ClawMetry or any of my open-source work saves you time, consider sponsoring. It keeps the OSS core free, funds the cloud infra I run, and lets me ship more.

_{⚡ Ship beats perfect. First-principles core, no bloat.}