Turning AI capabilities — LLMs, agents, document intelligence — into reliable, production-grade systems. I care about clear technical thinking, resilient architecture, and secure infrastructure as much as shipping fast.
Senior Software Engineer based in Florianópolis, Brazil, focused on back-end development, system architecture, and applied AI. My day-to-day spans AI-powered data/document pipelines, confidential computing (TEE-based secure infrastructure), and cloud-native backend systems — with a background in blockchain and tokenization that still informs how I think about trust, verification, and distributed systems.
I've moved from writing smart contracts to designing AI-native platforms — but the core hasn't changed: understand the problem deeply, design the system, ship it into production.
- Systems thinking first — architecture decisions before code
- Clean Architecture & Domain-Driven Design
- Security-first infrastructure — confidential computing, least-privilege by default
- AI as a production tool, not a demo — reliability, observability, and real business fit
- Event-driven design for loosely coupled, scalable services
| Area | Stack |
|---|---|
| Backend | NestJS · Next.js · Express · GraphQL · PostgreSQL (+ Prisma) · Redis · Event-driven architectures |
| Applied AI | Python · Claude & Gemini API integration · LangChain · RAG pipelines · pgvector · Document intelligence / OCR pipelines |
| Confidential Computing | GCP Confidential VMs · AMD SEV-SNP · Vault Transit · TEE-based secure AI workloads |
| Cloud & Infra | GCP (Cloud Run, Cloud SQL, Cloud Build, IAM) · Docker · CI/CD |
| Web3 / Solidity (background) | ERC-20/721/1155 · Hardhat · Foundry · OpenZeppelin · Multi-chain (Gnosis, Polygon, Base) |
| Quality | Jest · Vitest · automated testing pipelines |
- AI-powered document intelligence pipelines — extraction and structuring of technical/industrial documentation at scale for enterprise clients.
- Confidential AI infrastructure — secure data processing on GCP Confidential VMs (AMD SEV-SNP) for regulated, privacy-sensitive workloads.
- Multi-chain tokenization platforms — on-chain compliance flows (TREX/OnchainID concepts), wallet infrastructure, and KYC integration for real-asset offerings.
- AI-assisted engineering practice — building and mentoring around AI-assisted development workflows (Claude Code, Cursor) across a growing engineering team.
- Interpretability and mechanistic understanding of LLMs.
- Confidential computing patterns for multi-tenant AI workloads.
- Systems-level thinking as a durable edge as AI commoditizes routine execution.
