ML engineer for physical systems. I build machine-learning models that understand the machines and structures they monitor, and I ship them.
Mechanical Engineering graduate, University of Ilorin (4.09/5). I combine physics (FEM, degradation models, signal processing) with ML (deep learning, anomaly detection, surrogate models) to solve predictive-maintenance, structural-health and digital-twin problems.
| Project | Result |
|---|---|
| BeamTwin | FEM digital twin + physics-hybrid neural surrogate for beam damage detection. ~50× lower error than a plain network. FastAPI, Docker, CI, live demo |
| Filter remaining-useful-life | Physics-informed RUL from right-censored data: 58% lower error than gradient boosting; maintenance rule cuts cost 43% |
| Digital-twin anomaly detection | Detects 80% of unseen anomalies at 0.09 false alarms/day and names the root-cause sensor 98% of the time |
| KW51 bridge monitoring | Separates temperature from structural change on a real railway bridge; measures the retrofit's effect on natural frequencies (up to +2.1%) to within ±0.08 percentage points |
| Concrete crack segmentation | U-Net crack detection + width measurement; found a 79% train/test leak in a public benchmark |
| Heat-loss surrogate | Neural surrogate for optimising thermal insulation |
ML & data: Python, PyTorch, TensorFlow/Keras, scikit-learn, pandas, NumPy, SciPy Engineering: FEM, signal processing, Kalman filtering, time-series modelling Deployment: FastAPI, Docker, GitHub Actions, Render Currently learning: SQL & PostgreSQL, MLflow, cloud (AWS/GCP), LLMs & RAG