• Build applied AI systems that turn messy data into usable signals
• Work at the intersection of neuroscience, health data, and software
• Design backend-heavy systems with clean structure and real output
• Automate workflows that shouldn’t require human effort

I focus on clean architecture, fast iteration, and building things that actually get used.

Built a 3D deep learning pipeline to classify aneurysm vs. healthy vessel segments from volumetric brain MRA scans.
Implemented a custom 3D ResNet-18 with Squeeze-and-Excitation attention to capture subtle vascular morphology in small, imbalanced datasets (~1.3k samples).

Modeled the relationship between physiological stress and sleep latency using wearable data.
Engineered rolling biological baselines and Z-score features to capture deviation from personal norms, revealing non-linear stress thresholds missed by standard correlation methods.

Developing a system to predict personalized sleep windows using historical sleep behavior, daily activity, and stress signals.
Combines constraint-based modeling with time-series feature engineering to dynamically shift recommended sleep timing based on intraday physiological load.

Built a polyglot, microservices-based web scraping and analysis platform designed for scalable data collection and real-time insight generation.
Orchestrates Go-based concurrent scraping, Python NLP/ML pipelines, and Node.js browser automation behind a unified API layer, with a hybrid MongoDB/PostgreSQL data architecture.

If you want to chat, please reach out.

ethan4lobo@gmail.com
linkedin.com/in/ethanlobo