MSc Engineering Physics @ University of Coimbra
Building intelligent systems at the intersection of physics, control theory, and reinforcement learning — from research notebook to production VPS.
- 🧠 MSc Thesis — Multi-Agent Reinforcement Learning (G2ANet) for autonomous control of wastewater treatment plants using BSM2/Simulink
- 🧪 rl-sweep — 120-run PPO vs DQN hyperparameter study across 3 Gymnasium environments; discovered and root-caused a policy-collapse failure mode
- 🌾 Grain Sentinel — STL + rolling MAD anomaly detector for industrial temperature data, deployed unattended on a $6/mo VPS via cron
Languages & Frameworks
Areas
Reinforcement Learning Multi-Agent Systems Hyperparameter Studies Time-Series Analysis Anomaly Detection Control Theory VPS Deployment
| Project | Description | Stack |
|---|---|---|
| bsm2-marl | MARL for wastewater treatment plant control — MSc Thesis | Python · PyTorch · MATLAB/Simulink |
| rl-sweep | 120-run PPO vs DQN sweep across 3 Gymnasium envs — discovered & explained a policy-collapse failure mode | Python · Stable-Baselines3 · Gymnasium · multiprocessing |
| grain-sentinel | STL anomaly detector deployed on a $6/mo VPS — 88% recall, 52% precision | Python · pandas · statsmodels · Linux/cron |
When I'm not training neural networks, I'm training myself — competing in track & field for Académica de Coimbra.
Open to Remote or Hybrid ML/AI Engineer roles, plus freelance work in ML/AI and Python. Feel free to reach out!