About
I’m a staff engineer focused on recommendation and search systems. I currently build personalization infrastructure at scale, working with embeddings, data pipelines, and vector search architectures.
This blog is where I document what I run into in practice: benchmarks I’ve run, mistakes I’ve made, and architecture decisions that don’t fit in any official documentation.
What you’ll find here
- Search & Retrieval — vector search, ANN, common embedding mistakes, how Elasticsearch and OpenSearch really behave in production
- Embeddings & Vector Databases — pgvector, Qdrant, Milvus, comparative benchmarks, pitfalls in managing embeddings
- Recommender Systems — production recommendation architecture, cold start, evaluation that matters
- Data Infrastructure — PostgreSQL, MongoDB, migration decisions, what happens at scale
About me
Software engineer with experience in backend, data, and ML systems. I write in Portuguese first. Some articles are translated into English and French.
→ GitHub