AI Research Engineer · BTG Pactual · São Paulo, Brazil
Scientific Machine Learning and explainable AI for dynamical systems under uncertainty.
I work at the intersection of research and production — building interpretable models that extract governing structure from data, detect regime transitions, and quantify uncertainty in ways that are both mathematically grounded and operationally useful.
Core axis: SciML · xAI · dynamical systems · probabilistic inference
Lexis — Regime discovery in dynamical systems
BOCPD-based change point detection + SINDy sparse regression for governing equation recovery.
Pareto-optimal model selection (accuracy vs. complexity). Applied to infrastructure monitoring
— detected system degradation signals ahead of failure events.
HSP — Hidden Survival Paths (in progress)
Probabilistic estimator of local survivability under perturbation in dynamical systems.
Formalizes basin persistence via phase space geometry + Bayesian uncertainty modeling.
Targeting 2026 publication.
MIDAS
Causal inference framework for financial dynamical systems.
Graph Neural Networks + control-theoretic modeling over dynamic graphs.
Research-grade architecture. JAX / Equinox backend.
nova-selachiia — Ecological modeling under uncertainty
Neural State Space Models → Deep Markov Models with Monte Carlo sampling.
Rare event modeling, survival analysis, and counterfactual reasoning.
Research: JAX · Equinox · Diffrax · Optax · PyTorch · PyG · SINDy · Neural SDEs
LLM Systems: LangChain · LangGraph · Agno · LiteLLM · Langfuse · RAG · MCP
Engineering: Python · R · Docker · Kubernetes · MLflow · ONNX · Neo4j · FAISS
Building toward a research-oriented MSc (Europe) with focus on SciML and interpretable dynamical systems. Open to collaborations at the boundary of scientific computing, probabilistic modeling, and real-world complex systems.
EU Citizen · Open to relocation