I built an on-device hybrid search engine that combines BM25 and vector retrieval with Reciprocal Rank Fusion. Reranking metrics suggested a learned linear fusion model would outperform RRF, but end-to-end evaluation showed otherwise. This article explains why the model matched baseline behavior and what to improve next.
A short bugfix session became a practical example of agent supervision. The agent produced valid code quickly, but it focused on symptoms first. Progress came from brief human questions that redirected work toward root cause, accurate impact framing, and a cleaner final production change set for users.
aws s3 sync was running at 60 KB/s on a link that should reach around 80 Mbps. The issue was not one bug but several: DFS channel pauses, Apple roaming behavior, LTE bufferbloat, and smaller configuration problems. This article walks through the complete fixes on RouterOS v7.
Intuitive prompt engineering - often called vibe coding - promises a flow state for software engineers, but the reality is often a repetitive loop of review and correction. We’ve traded writing code for supervising agents that write code. The question is how to make that trade worthwhile.