I'm Erik Cohen, a hands-on AI engineer
I build AI agents, the retrieval behind them, and the evals that show whether a change helped. Voice is one place I've done it. The method is the same for any LLM system.
11+ years shipping software. MS CS (ML), Georgia Tech, completing 2026.
What I've built
From 2021 to 2025 I built production AI for scientists at BenchSci and owned evaluation end to end. Read the case study.
Independent contractor (Cohen AI Consulting LLC), AI/ML engineering for a US healthcare-technology company, Jan–Oct 2026. See one call, stage by stage.
How I work
I do the engineering myself. I write the code, ship it and measure it.
I orchestrate AI agents to do more as one engineer. I measure them the way I measure any AI system: graded runs, per-model scorecards, and audits by a second model.
Before a change ships, there is a test that says whether it made the system better or worse.
Work with me
Tell me what you're building and what's failing. In 30 minutes we'll see whether an engagement fits.
Recruiters: LinkedIn is fastest. Code is on GitHub.