At Caat_ingá Observatory (OCA), I own product across four lines, including a regional data platform for SUDENE and an environmental atlas with FAO. For SUDENE’s automated municipal report, I set up an editorial review between engineering and the client’s data team; within four weeks, eight themes passed quality checks, up from two.
I sped up the report generator in Python and PostgreSQL, bringing repeat report returns from 8.4 seconds to 31 milliseconds. I profiled each stage with EXPLAIN ANALYZE and moved heavy queries to materialized views, while caching generated content and pre-rendering maps for all 2,074 municipalities.
Before OCA, I worked on distributed-systems tooling in Go and React and published research on evaluating LLM agents. I also shipped Mô, a household chores app, as a solo developer, and my research on workload migration in federated Kubernetes environments has been published and accepted for presentation.
