Senior Solutions Architect at MongoDB · Data & AI platforms
I design and build production-minded proofs of value at the intersection of operational data, search, AI, security, and real-time systems. My work turns an architecture decision into something stakeholders can inspect, run, measure, and challenge.
Public-facing overviews and code are written in English. PoV interfaces and presentation scripts may remain in Brazilian Portuguese when they are designed for local customer conversations.
| Project | What it demonstrates |
|---|---|
| Torre — Atlas Control Plane | Fleet health, FinOps, scaling, Performance Advisor, observability, and an AI assistant grounded in live Atlas metrics |
| MongoDB to Apache Iceberg | Atlas Stream Processing CDC into S3/Iceberg, including updates, deletes, schema evolution, time travel, and measured propagation latency |
| Queryable Encryption | Equality and range queries over randomized ciphertext, with application and DBA views shown side by side |
| RAG on MongoDB Atlas | Hybrid retrieval with Vector Search and BM25, RRF fusion, reranking, citations, tenant isolation, and default-deny ACLs |
| Economic Group Graph | $graphLookup, fuzzy entity resolution, concentration analysis, visibility hierarchies, and measured graph workloads |
| Multi-Agent on MongoDB | MongoDB Atlas as both the data plane and coordination plane for stateful, observable, policy-constrained agents |
- MongoDB data modeling, performance, sizing, and distributed architecture
- Atlas Search, Vector Search, hybrid retrieval, and AI agent systems
- Real-time processing, change streams, CDC, and lakehouse integration
- Encryption in use, tenant isolation, guardrails, and auditable write paths
- Observability, resilience testing, FinOps, and evidence-based trade-offs
- React/FastAPI applications that make architecture inspectable in a live PoV
- Start from the decision the customer needs to make.
- Model the failure modes and security boundaries before the happy path.
- Use synthetic or anonymized data and keep credentials outside the repository.
- Measure behavior against real infrastructure whenever the claim depends on it.
- Document limitations and trade-offs explicitly; a demo is evidence, not a production guarantee.
- São Paulo, Brazil
Opinions and experiments published here are my own. MongoDB and MongoDB Atlas are trademarks of MongoDB, Inc.