Marawan Yakout
Computer scientist @ University of London · Machine Learning Research Intern @ DeepKapha Labs
About
A foundational researcher with undergoing Computer Science BSc - University of London with a focus on AI, Machine Learning, Data Science and DeepLearning. Passionate about research, reproducibility and high‑quality open‑source implementations.
What I work on
- Deep learning and generative models (e.g., diffusion for sky/cloud understanding).
- High‑performance C++ and Python systems for trading and large‑scale experiments.
- Applied AI research, model evaluation, and educational content.
Recent Work
-
SERWED
Diffusion Model • Satellite Imagery • Conditional Generation
GitHub: https://github.com/MarawanYakout/SERWEDA context-conditioned Denoising Diffusion Probabilistic Model (DDPM) for generating and analyzing satellite cloud imagery. Trained on a NASA dataset containing high-resolution images with associated timestamps, wind-speed annotations, and learned embeddings. Supports conditional generation (e.g., given wind speed or timestamp), robustness experiments, and rapid prototyping on small 16×16 images with straightforward scaling to larger resolutions.
Project Status: Completed - on arXiv / under review
Paper and full quantitative results link soon to be added
Contact
- GitHub: github.com/MarawanYakout
- LinkedIn: linkedin.com/in/marawanyakout
- YouTube: youtube.com/@centrified
- X/Twitter: twitter.com/MarawanYakout
- Email: [email protected]
- Email: [email protected]