I am a data scientist with experience leading AI research projects: reducing LLM evaluation costs by 35%, increasing recall by 50% in a RAG system, and fine-tuning a 3.8B parameter model to outperform GPT-4 Turbo in tool use. I am passionate about applying cutting-edge research to generate impactful results. I have ML engineering skills from co-leading technical work on an ML bias/explainability Python package at PwC (exp. return of £m’s) and building RAG products & pipelines at Morningstar. My MPhil dissertation at the University of Cambridge applied ML techniques to predict taxi driver behaviour.