About
Brazil's market data is public, but it's scattered across the securities regulator (CVM), the central bank (BCB), the exchange (B3), ANBIMA, trustees and rating agencies — in formats nobody wants to touch. Most of my work is turning that into one documented database an analyst — or an AI agent — can query directly, and building what runs on top of it: issuer rating models, inflation forecasting, fixed income pricing, and the apps the team uses day to day.
It's built and maintained with heavy use of AI coding agents, mainly Claude Code. I studied economics; most of what I do now is engineering.
What I build
- A catalog of Brazil's public financial data — 200+ automated routines running daily, in place of paid data vendors.
- A 2TB+ database, documented field by field, so analysts and AI agents can query it directly.
- Proprietary rating models for corporate, financial institution and fund issuers.
- An inflation forecasting model that has ranked among Bloomberg's top IPCA forecasters.
- A fixed income pricing engine, an investor portal, and the internal apps around them.
Writing
- Curva de juros: um conto de três fatores — PCA aplicado à estrutura a termo de juros brasileira.