Projects
AI integrations, MCP tools, and backend systems I worked on: the problem, my part in it, the decisions and the result. From AWS infrastructure to tools running in production. Internal systems are described without code or confidential data.
MCP for AI: from AWS infrastructure to production tools
End-to-end delivery of an MCP server for AI assistants, from AWS infrastructure and deployment to tool design, implementation, and observability in production.
- Node.js
- TypeScript
- Model Context Protocol
- OAuth 2.0
- AWS Lambda
Public API platform for a truck service network
Contract-first endpoints, synchronized documentation, and database-focused performance investigations for a public REST API.
- Ruby on Rails
- PostgreSQL
- dry-rb
- JSON:API
- OpenAPI
Traffic-light control for São Paulo and wagon tracking for the rail network
Two real-time systems at a small consultancy, one talking to intersections over MQTT and the other mapping freight wagons through a SOAP integration, built with TDD from the first commit.
- Java
- Ruby on Rails
- Angular
- MQTT
- MongoDB
Profile-based authorization for a billing platform
A permission model flexible enough for finance teams to define their own roles, built as a consultant for Ordway.
- Ruby on Rails
- React
- PostgreSQL
- Redis
- Heroku