Series · 5 parts
Agentic AI Stress Suite
Nine planned experiments, four published so far, on where RAG-to-agent stacks break: retrieval, evaluation, agents, LLM ops, security, scaling, and custom embeddings. One harness, one corpus, negative results and corrections published.
Start with part 1: Do HyDE and Query Decomposition Improve RAG Retrieval?parts5
first post2026-09-20
latest2026-10-25
total read~112 min
other topicsAllBackpressure & Load ControlProject LoomStructured ConcurrencyTail Latency & System Behavior