Sam Bretz

Sam Bretz

Founding AI engineer at Metergraph, helping teams cut LLM spend without losing quality.

Metergraph reads production traces from tools like Langfuse, LangSmith, Braintrust and OpenTelemetry, classifies workloads, and replays them on candidate models against evals to produce evidence-backed model-swap recommendations. Before that I built the core of AG2's agent platform, and before that worked on platform, data and distributed systems, from surgical video pipelines to enterprise identity graphs. I write here about engineering, AI tooling and whatever else I am curious about.

Now

Before

Writing

First posts are on the way.