Michael Bennett
AI Systems Builder | Security | Automation
New to Contra
Michael is ready for their next project!
I extended the open-source Paperclip project into a multi-company AI operations control plane with real cost and governance controls. Every agent operates under per-project budgets with automatic hard-stops. Board governance includes approval gates for hiring agents and strategy changes. Built with TypeScript, React, PostgreSQL, Drizzle ORM, and Express. Open-source contributions are public; proprietary extensions available under NDA.
0
26
TheBinMap is a 528-page nationwide directory of liquidation bin stores that I built from zero to live in under 10 weeks. It uses an automated data pipeline — Google Places discovery, AI enrichment, and static page generation — with zero server costs. It ranks on competitive local search terms and has multiple monetization paths. Built with Astro, Python, SQLite, and DeepSeek LLM integration. Live at thebinmap.com (http://thebinmap.com).
0
33
I identify security and protocol gaps in major open-source infrastructure and ship fixes upstream. Contribution to Coinbase's x402 micropayment protocol (merged PR): added a security API built with TypeScript. Contribution to OQTOPUS quantum ecosystem (PR #79): built a zero-dependency Python attestation module using standard-library SHA-256 cryptography that detects tampered quantum hardware jobs. Full details at mbennett-labs.github.io/contributions.html (http://mbennett-labs.github.io/contributions.html).
0
36
QuantumShield Core — Governed AI Security Operations Platform.
Problem: Organizations using AI for security cannot prove, months later, that a human actually reviewed and governed AI output. When audits arrive, they have timestamps but no evidence of judgment.
What I built: A production governance platform where every AI-assisted security decision goes through human review with mandatory written reasoning. Every decision is recorded in an immutable, append-only log that can be replayed step-by-step. The system enforces strict rules: no autonomous governance action, no hidden fallbacks, no record deletion. I built the operator interface, validated structural integrity, and documented the entire governance philosophy.
Key capabilities: human-in-the-loop approval gates, decision replay, lineage visualization, governance scenario training lab.
Tools: Python, JavaScript, PostgreSQL, Security Architecture, Governance Design.
Learn more: https://quantumshieldlabs.dev/
0
41