Hi! I'm Emmanuel, an applied machine learning researcher focused on Generative AI.
DPhil Engineering Science, University of Oxford (viva pending). I build and evaluate end-to-end ML systems on large, noisy real-world datasets, with first-authored publications and industry collaboration.
Portfolio
Publications
- Loading publications...
Projects
Hackathons
- Face2Visceral: Visceral Fat Prediction from Facial Images [6hr Hackathon, 3rd place]. Predict visceral fat ratio from facial imagery using a PyTorch Lightning model and a lightweight FastAPI inference service.
- HandRiGht: Google hackathon project for handwriting support in people with cerebral palsy, delivered as a working prototype in under 24 hours (2nd place out of 12 teams).
Software
- ECIQC: Tool developed between SABS:R3 (University of Oxford) and GE HealthCare to automatically validate DICOM images used for medical research.
- sciProject-MCMC: Animates two common Markov Chain Monte Carlo methods (Random Walk Metropolis and Adaptive Covariance MC) on bivariate normal distributions.
Skills
Python, MATLAB, C++; PyTorch, HuggingFace, NumPy, SciPy, Scikit-Learn; generative modelling, diffusion, flow matching, representation learning, contrastive learning, and robust evaluation on large-scale real-world datasets.
Production & Deployment
- Reduced computational cost by 30–50% in a production reservoir simulator by integrating machine learning models during my Schlumberger internship.
- Built and maintained robust data tooling with CI/CD and automated tests for DICOM quality control workflows in collaboration with GE HealthCare.
- Designed and evaluated statistical/ML pipelines on 4M noisy records from 208,948 admissions in a large-scale clinical dataset study.
My work spans generative modelling research, applied machine learning engineering, and collaborative projects across academia and industry. Recent work includes diffusion and flow matching methods, synthetic data generation for downstream tasks, and robust model evaluation in data-constrained settings.
MICCAI 2023: Poster Presentation
MICCAI 2025: Daejeon, South Korea
MICCAI 2025: Poster Presentation
ASMUS-MICCAI 2025: Best Paper Award
About Me
I am a DPhil Engineering Science candidate (viva pending) at Oxford with a research focus on generative modelling for high-dimensional medical image and video data. My work covers diffusion models, flow matching, representation learning, and foundation-model adaptation, with emphasis on robustness, scalability, and responsible evaluation. Across research and internships, I have developed end-to-end ML workflows from data curation and model training to analysis and deployment-oriented optimisation, including projects with GE HealthCare and Schlumberger.