Big Tech, Scale-Up, or Startup? The Brazilian SWE Landscape
5 min read
Brazil has all three now. But they offer very different things depending on where you are in your career. Here is what each one actually looks like from the inside.
Blog
Writing about what I learn — data infrastructure, AI systems, and the engineering decisions behind them.
AI Doesn’t Change Architecture — It Exposes Bad Architecture
7 min read
Teams blame AI for making their systems complex. But most of the time, the complexity was already there. AI just made it impossible to ignore.
The Architecture Tax of AI Features
6 min read
Adding AI to a product is never just an API call. There is a hidden tax in latency, cost, observability, and failure handling that most teams discover too late.
My New Area: Semantic Search and Content Understanding
5 min read
I recently moved into the Semantic Search and Content Understanding space at Microsoft. Here is what this area is about, why it matters for the age of LLMs, and how concepts like vectorization and RAG power products like Copilot, ChatGPT, Claude, and Gemini.
Big Tech or Startup? What Nobody Tells You Before You Switch
8 min read
I started at Kunumi, a small AI startup in Brazil, and moved to Microsoft. Both had things I did not expect. Here is what I learned about what each environment actually gives you.
LLM Fundamentals That Actually Matter in Production
6 min read
Most LLM explainers focus on transformer architecture. Engineers building real products need to understand something different: the practical fundamentals that determine whether your system actually works.
The Product Won’t Save You
4 min read
Engineers stay in broken environments because the product is interesting. But a great product on a bad team is a trap that costs you more than you think.
Data Quality Is a User Behavior Problem
7 min read
Most data quality frameworks validate structure and values. The hardest regressions come from somewhere else entirely: the humans generating the signals changed how they behave.