Evals before vibes
Every AI feature gets a test set, a metric, and a baseline before it gets a demo.
Skills
A working map of what I reach for — from model orchestration to pixels to infrastructure. Depth where it matters, breadth where it helps.
Categories
Drag to rotate · hover to inspect
Proficiency
Retrieval, agents, evals and the plumbing that keeps them honest in production.
Interfaces that feel fast, look considered and stay accessible.
Typed contracts, streaming responses, and data models that age well.
From event contracts to forecasts — the analytics side of my USC degree, applied.
Shipping into customer environments means being comfortable in all of them.
The non-code half of the job: discovery, integration, enablement and trust.
How I work
Every AI feature gets a test set, a metric, and a baseline before it gets a demo.
Forward deployment means the spec lives in the customer's workflow, not in a doc. I go find it.
Schemas at every seam — API, DB, model output — so the system fails loudly and early.
Tracing, logging and feedback loops are part of v1, not a follow-up ticket.
Daily toolbelt