Now
Updated Mar 20, 2025. A living log of what I’m reading, learning, writing, and building.
Reading
- Evaluating RAG Systems: Beyond Accuracy — good rubric for business UX
- Patterns for Agent Workflows — guardrails + fallbacks
- Prompt Compression and Context Windows — pragmatic, not hype
Learning
- RAG evaluation harness (lightweight) — precision/recall style + UX rubrics
- Small models + adapters (LoRA) — cost‑aware experiments
Writing
- Pricing prompts like products — drafting
- From prompts to SOPs — case study
Building
RAG for Small Businesses
- Added chunking strategy switcher (tokens vs sentences)
- Implemented eval baseline + error notebook
- Reduced cost by ~38% via indexing tweak
Research Companion
- New compare view
- Citations clean‑up