AI engineer building agents for user growth.
I build the infrastructure behind measurable growth: agent workflows, short links, event tracking, experiments, and the systems that verify the outcome.
- event-contracts — an offline Python CLI and typed API for catching analytics schema and privacy drift before it reaches production. v0.1.0 ships 56 tests and clean-wheel CI across Python 3.12–3.14.
- growth-skills — installable, testable Agent Skills for SEO surface audits, trend discovery, measurement contracts, and privacy-safe campaign links. v0.2.0 ships four deterministic packages backed by 47 tests and cross-platform CI.
- auto-redirect — config-driven short links and privacy-minimized click attribution on Cloudflare Workers. v0.1.0 ships typed redirect rules, deterministic query precedence, and 28 CI-backed tests.
- Vercel AI SDK contribution — an open implementation PR for Zod 4 record conversion so JSON Schema preserves the record value schema instead of silently rejecting every key.
- AI agents with explicit success criteria, verification, and human approval
- Short-link and redirect systems that preserve attribution
- Event pipelines that turn product behavior into reliable decisions
- Growth tooling in Python, TypeScript, JavaScript, and Go
I care about one practical question: can the system prove that the work actually happened?

