6 articles tagged with MLOps
A walkthrough of taking a narrow task off an expensive frontier model and onto a small fine-tuned model that costs a fraction as much to run, using the frontier model as the teacher.
A survey of the layers that make up a real agentic system today — orchestration, memory, tools, evaluation, guardrails, observability — and the specific ways teams get each one wrong.
Open-weight models went from 'interesting but behind' to a default option on my shortlist for real workloads. Here's what actually shifted, and where the API-only providers still win.
Unit tests don't work on non-deterministic outputs. Here's how to build an evaluation pipeline that catches regressions in an LLM app the same way a test suite catches them in ordinary code.