How I’m combining keyword and semantic search in Trama for Obsidian to retrieve notes with less friction and discover connections worth linking.
Andrea Casarin
A collection of 92 posts
Why I’m Building Trama for Obsidian
I use Obsidian daily for diary entries, reviews, and finding old notes. I’m building Trama around that workflow: keyword + semantic search, related notes, and AI chat. Privacy, mobile use, and a clean interface matter to me.
Why I Self-Host: Analytics with Umami
Why I self-host Umami analytics on Coolify for personal projects: useful privacy-conscious insights, predictable costs, and infrastructure I’m comfortable maintaining.
Fantastic CSPs and where to find them
A practical guide to Content Security Policy: what browsers block, how CSP limits XSS damage, and how to test a policy without breaking your site.
You Don’t Need a Vector DB: Running hybrid search in production
Production notes on running hybrid search in PostgreSQL: ANN indexes, HNSW vs IVFFlat, trigram indexing, memory, and query limits.
You Don’t Need a Vector DB: Ranking hybrid search in Postgres
How I ranked hybrid search in PostgreSQL by combining `pgvector` and `pg_trgm` with Reciprocal Rank Fusion, without mixing incompatible scores.
You Don’t Need a Vector DB: Setting up hybrid search in Postgres
A practical look at setting up hybrid search in PostgreSQL with `pgvector` and `pg_trgm`, using a simple schema and a small ingestion flow.
Green-blue Deployments with Docker and Ansible
Using Ansible to deploy a small fleet of containers via Docker Compose, green-blue zero downtime stuff and takeaways