Articles
Everything I learned the expensive way.
Long-form technical articles on retrieval, multimodal embeddings, vector databases and context engineering.
CodeBot: context engineering for coding agents
How to build semantic code search for a repository with Graph RAG: hybrid search, neighbor expansion, community detection and so on.
Dec 28, 2025 · 9 min read
ImageDB
A solution to search over millions of figures extracted from arXiv papers, and how to deploy it for $13/month on a VPS.
Oct 18, 2025 · 13 min read
RAG with ColPali
A full guide to multimodal retrieval with ColPali: late interaction, token pooling, approximate search, and the trade-offs of putting it in production RAG.
May 24, 2025 · 9 min read
PostgreSQL with pgvector
Searching millions of vectors without leaving PostgreSQL: index choice, recall tuning, and the places pgvector quietly stops being fast.
Also published
On Medium
Feb 26, 2025 · Medium
How to count the Transformer’s number of parameters
Working out, layer by layer, where every parameter in a transformer actually lives.
Feb 24, 2025 · Medium
A story of clean code: an ML use case — Part II
Refactoring a real training pipeline with design patterns that survive the second experiment.
Feb 8, 2025 · Medium
A story of clean code: an ML use case — Part I
Why most ML codebases rot, and what a clean one looks like before the deadline hits.
Got a problem that looks like one of these?
If any of this is the thing you are stuck on, that is usually a good sign we should talk.