Vector engineering for agents
A deep dive on the retrieval layer behind RAG systems and AI agents - written for engineers who have to make these systems work in production, not just in demos.
What it covers
- Embeddings and approximate nearest-neighbor search trade-offs
- Hybrid retrieval: combining dense, sparse, and structured signals
- Agentic patterns and agent memory design
- Evaluation and production operations for retrieval pipelines
Why it exists
Retrieval quality is where most agent systems quietly fail. The material collects the decisions that actually matter - index choice, chunking, reranking, memory, and evals - in one place, at the depth needed to defend them in a design review.