First Year MPhil/PhD student at UCL Institute of Health Informatics, developing machine learning frameworks for rare disease research. First Class Neuroscience BSc Hons from the University of Bristol. Recipient of the 2024 Snowdon Master's Scholarship and member of the Windsor Fellowship.
May 2025 - Present
- Developing ML frameworks for understanding EDS-POTS-MCAS disease relationships
- Large-scale web scraping of StuffThatWorks.com (2.9M members, 110M data points)
- Implementing multi-modal GARNN architectures for disease trajectory prediction
- Winner of UCL Digital Health AI Hackathon (April 2025)
2021 - 2024
- First Class Honours with PLUS Award
- Dissertation: "Ivabradine and the Risk of Torsades de Pointes" (graded 84, excellent 1st class)
Research Scientist Intern @ Advai (2025) - Mechanistic interpretability research on bias in medical LLMs using Sparse Autoencoders Freelance ML Consultant (2025-Present) - Building AI systems for strategic decision-making with historical case analysis Tech Intern @ Elsevier (2024) - Developed bias mitigation tools for Clinical Key AI and RAG systems for medical literature Computational Neuroscientist @ Neuro-Bio (2023-2024) - ML/AI drug discovery, achieving 99.98% analysis accuracy and £650k+ cost savings with custom ALzheimer's drug design
Programming Languages
- Python (PyTorch, Pandas, NumPy, SciPy, Django, scikit-learn)
- JavaScript (React, Node.js, Full-stack development)
- R (Statistical analysis, ggplot2, tidyverse)
- MATLAB
- Version Control: Git, GitHub
Machine Learning & AI
- BERT/BioBERT, RAG models, Knowledge Graphs
- Graph Neural Networks (GNNs), GARNN
- Mechanistic Interpretability: Sparse Autoencoders, feature clamping
- XGBoost, Random Forest, SMOTE
Scientific Computing
- Molecular Modeling: AutoDock Vina, RoseTTAFold
- Cheminformatics: RDKit, MOE, QSite
- Web Scraping: BeautifulSoup, Selenium
- Snowdon Master's Scholarship (2024) - Scholarship for exceptional students to enable further study in their field of choice & promote inclusivity and change as leaders in their subject area
- Digital Health AI Hackathon Winner (2025) - UCL Faculty competition, developed ScleroDx diagnostic tool
- First Place - Bristol Student Research Festival, Sustainability and Wellbeing Category
- PLUS Award - University of Bristol, for 70+ hours of professional development
- GARNN architecture achieving 79% accuracy on EUSTAR dataset
- Addresses health equity in rare disease diagnosis
- Live deployment: https://scleroderma-api.onrender.com/
- GitHub Repository
- BioBERT-based system for automated drug safety analysis
- Open-access web tool for torsadogenic medication reviews
- Manuscript in preparation
- GitHub Repository
- Leading research on hypermobility/EDS and neurodiversity links
- Part of Disability Action Research Kollective
- Building consensus on diagnostic overlaps
- 2024 - Manuscript in preparation: "Ivabradine and the Risk of Torsades de Pointes: A Case-Based Analysis"
- 2023 - Bristol Student Research Festival (1st Place): "The Neurobiology of Ethical Decision Making"
English (Native) • Farsi (Fluent) • French (Fluent) • Italian (Conversational) • Turkish (Conversational)
- 📧 Email: amelia.ghanea-hercock.24@ucl.ac.uk