Full Stack Engineer building AI products and multi-agent systems. Florianópolis, Brazil.
I'm the cofounder of RBBT, where we build multi-agent systems for real operations, and of Fomenta, an AI platform that helps researchers find and access funding. Most of my work sits where the AI layer meets production: data pipelines, orchestration and guardrails. I build end to end, from interface to infrastructure, and I like problems where the hard part is everything around the model.
- Building agent systems at RBBT and scaling Fomenta.
- Thinking about how autonomous agents earn trust: observability, clear boundaries, and knowing when a person should take over the decision.
RBBT — Sales · Cofounder
A multi-agent AI sales system that qualifies customers, builds carts and closes orders over WhatsApp and web chat, handing off to a person only when needed. Four least-privilege agents (routing, selling, checkout, post-sale), a platform-agnostic core with Shopify and Nuvemshop adapters, fail-closed human escalation and continuous agent evaluation.
TypeScript Anthropic Claude PostgreSQL WhatsApp Cloud API Shopify Nuvemshop
Fomenta · Cofounder & CTO · Programa Nascer (FAPESC)
An end-to-end SaaS that finds public research funding for Brazilian researchers and matches each opportunity to the researcher's profile, with architecture from ingestion to payments. A nine-Lambda, event-driven ingestion pipeline over SQS with LLM-based parsing and normalization; dead-letter queues and CloudWatch alarms for failure containment; an async semantic matching service with prompt-versioned caching; Stripe subscriptions and usage quotas.
Next.js TypeScript AWS Lambda SQS DynamoDB Supabase LangChain Puppeteer Stripe
SwellGuide · paused
A daily newsletter that turned raw weather data into plain-language surf forecasts. I designed, launched and grew it to 50+ daily active users, with a modular design for adding surf spots and an automated Node.js + LangChain workflow that ran the data through an LLM and dispatched the campaigns. How I built it →
Node.js LangChain AWS Tailwind CSS
- Agents Are the Easy Part: The Harness Behind an AI Sales Team — Nov 2026
- From Idea to Production: How I Built SwellGuide End-to-End — Jan 2026
- System Design Basics: Caching · Database Sharding
| Languages | TypeScript, JavaScript, Python |
| AI | Anthropic Claude, LangChain, multi-agent orchestration, RAG, evals |
| Front-end | React, Next.js, Tailwind CSS |
| Back-end & data | Node.js, PostgreSQL, DynamoDB, Supabase, MongoDB |
| Cloud | AWS (Lambda, SQS, EventBridge, DynamoDB, CloudWatch, SAM), Cloudflare Pages |
| Integrations | WhatsApp Cloud API, Shopify, Nuvemshop, Stripe, Puppeteer |
- Full-Stack Development at FIAP, after a background in Public Administration at UDESC. That is where my analytical, user-centered way of looking at problems comes from.
- Eight months living in Dublin, nine countries, 20+ cities.
- Florianópolis is an island with 42+ beaches, and surfing is how I reset.
Open to conversations about agents in production, AI product engineering and new collaborations. The fastest way is LinkedIn or my website.



