Full-stack engineer: Java/Spring Boot and React/TypeScript. Four years on a 20-engineer team at KBR, where I led the adoption of automated testing and built the team's first flow-analytics dashboards. I build software with AI agents every day, and I ship the evidence: tests, CI, evals, demo modes.
Portfolio: jacobv25.github.io · LinkedIn: jacob-valenzuela
| Project | What it is |
|---|---|
| ado-flow-metrics | Flow metrics for Azure DevOps teams: WIP and PR aging, throughput, cycle time, Monte Carlo forecasting. An "Explain with AI" feature backed by an 11-case eval harness with deterministic graders and an LLM judge. Vitest + Playwright, CI, demo mode. |
| mise | React Native cooking companion built around mise en place. A markdown-to-JSON recipe pipeline with parser contract tests. 72 Jest tests, CI. |
| speedhell | A browser bullet-hell with zero dependencies, built through a builder / blind-critic / deterministic-referee loop. Playable. |
| undertow | Android app that interrupts doomscrolling with a friction overlay (AccessibilityService). CI, releases, MIT. |
| pg-tuning-lab | PostgreSQL schema-design and query-tuning lab: every optimization measured before and after with EXPLAIN (ANALYZE, BUFFERS) on ~2M rows. |
Open to remote roles, and to relocating for the right team.

