Your Evals Have No Tests
15 of 19 escaped · 79% escape rate on the sample run
A green eval suite isn't evidence your agent works — it's the absence of evidence. Mutation testing shows you exactly how blind your evals really are.
Experiments
Posts built around a measured result: a stated claim, a fixed setup, the numbers with their sample size or interval, and a verdict. Null results count too.
2 experiments · 1 benchmark
Choosing a Vector Database: Benchmarks from My Own Workloads
Every vector DB benchmark was run on someone else's data. Benchmark your own vector count, dimensions, and — above all — your real filtering pattern.
Cutting LLM Inference Cost ~40% Without Losing Quality
~40% lower cost · NDCG@10 held at or above 0.75
Inference cost optimization is a measurement problem in disguise. Fix the quality metric first, then trim context, route models, and cache the stable prefix.