- 🎓 MS Data Science @ Harvard
- 🎓 BSc Politics and Data Science @ LSE
- 🔬 AI researcher @ ML Foundations Group, Harvard John A. Paulson School of Engineering and Applied Sciences
- 🔬 AI researcher @ AI4LIFE Lab, Harvard John A. Paulson School of Engineering and Applied Sciences
- 📊 Former research assistant @ LSE Data Science Institute
- 💼 Former intern data scientist @ The Tecsa Group
- 💬 Programming languages:
- Python (advanced)
- R (advanced)
- TypeScript (beginner)
- SQL
- Mechanistic interpretability
- AI Safety
- Knowledge Distillation.
- Surviving Harvard... 🤓
- Are Personas All You Need? Stress-Testing Persona Vectors as Alignment Tools: Evaluated the reliability of persona vectors as low-dimensional alignment tools for modelling behavioural traits.
- DataDetox: Agentic AI system that traces model lineage and training data by piecing together disparate data sources from Huggingface and arXiv, improving AI safety and model governance by increasing transparency, allowing ML practitioners to easily identify a model’s upstream dependencies and prevent any risks from propagating downstream.
- ChatLSE: Developing an LLM chat bot for LSE using Retrieval-Augmented Generation (RAG).
- Discordia: Built a machine learning framework to identify factors that drive MP rebellions in UK Parliament.
- What's for Dinner: Fridge management app that tracks expiring food and leverages AI agents to recommend recipes that include these items (Finalist for Palantir's 2024 Winter Fellowship).
- FPL-Moneyball: Model to optimise team construction in Fantasy Premier League.
- LSE Data Science Module Final Project: Using machine learning algorithm to predict the outcome of NBA Most Valuable Player (MVP) award.