Guardrails for RAG
Nemotron Safety Guard 8B in front of a 70B.
Jul 22, 202610 min read1
Command Palette
Search for a command to run...
Series
Depth First
Depth First is my newsletter on production AI engineering. Each issue goes deep on one problem: guardrails, evaluation, LLM gateways, retrieval and the infrastructure underneath, with the configs, numbers and trade-offs I hit building them.
API gateways for LLM calls
Fallbacks, cooldowns, caching, and observability.
Jul 27, 202611 min read
Evaluating your RAG pipeline
Error analysis, golden datasets, context recall, and faithfulness.
Aug 4, 202616 min read
Inside Cursor's Git Host
Git hosting on S3, ordered by one compare-and-swap.
Aug 24, 202611 min read