Evidence & causality
Connecting decisions to defensible evidence, explicit assumptions, and an understanding of what actions can actually change.
APPLIED SCIENTIST & TECHNICAL LEADER
NelsonGuimarães
Applied AI for real-world decisions.
I build intelligent systems that support real world decisions. At Amazon Japan, I lead applied AI work on logistics services, combining machine learning, causal inference, forecasting, and optimization to solve complex operational problems.
Over the past decade, my work has taken me through banking, consumer analytics, and logistics, with roles in Brazil, Switzerland, and Japan. I hold a master’s in Computer Science from Georgia Tech.
Research interests
When should we trust
an AI decision?
I’m exploring the foundations of reliable AI decision systems: how they reason under uncertainty, respect constraints, and recognize when to defer or seek verification.
Uncertainty & verification
Recognizing the limits of a system’s knowledge, checking its reasoning, and knowing when more evidence or human judgment is needed.
Constraints & reliability
Bringing optimization, feedback, and safeguards into AI systems that operate within real-world requirements.
Writing
Notes on making better decisions.
· 4 min read
When Can We Trust Intuition?
A conversation between research on judgment errors, intuitive expertise, and the conditions under which simple decision strategies work.
Selected experiment
Talking Plants
Hardware · Software · AI
A playful experiment in giving houseplants a voice.
An ESP32 sensor device, a Flutter app, and a cloud backend connect environmental measurements with a conversational interface.
The project brings together embedded hardware, cloud infrastructure, and AI to make plant care a little more tangible—and a little more fun.
Explore the projectContact
Get in touch.
For conversations about applied AI, reliable decision systems, or building useful things.
Connect on LinkedIn