Notes from shipping LLMs to prod.
Long-form engineering write-ups on the parts that are actually hard: grounding a model against a fixed fact set, designing loops that provably terminate, deciding what a system should do when it runs out of retries, and keeping the whole thing alive on a box you own.
Your AI Resume Builder Is Lying About You
I built ResumeOS to tailor a resume to any job description without inventing a single fact about you. This is the grounding architecture that makes that a guarantee instead of a hope — a judge model handed the whole profile as ground truth, a bounded verify-and-repair loop, and one rule: fail closed on correctness, fail open on cosmetics.
- LLM Engineering
- Guardrails
- FastAPI
- LaTeX
- Prompt Injection
- Self-hosting
Opens two tabs — the write-up and resume.thetan.in
More in progress
Next up: building the eval set that turns the 70% gate from a considered guess into a measured decision boundary.