Engineering notes — things I keep re-explaining, written down once.
- How to moderate user content at scale with batch LLM classification and a review queue
- Batch LLM APIs vs real-time calls: how much cheaper are async bulk jobs?
- One API key for OpenAI, Claude, and Gemini: picking a gateway with fallback routing
- Cheap embeddings and rerank for semantic search: how to compare cost per 1M tokens
- Cheap Bulk CSV Tagging With LLM APIs: Use a Batch Job, Not Per-Row Classification