About Me

I’m a senior machine learning engineer based in Cambridge, currently building production LLM systems at Luminance. My work focuses on training, evaluating, optimizing, and deploying domain-specific language models for legal workflows where accuracy, speed, reliability, and privacy matter.

Recently, I led the development of a proprietary legal LLM now used in production for extractive contract-analysis tasks. The work covered the full path from experimentation and data preparation through fine-tuning, evaluation, inference optimization, and deployment, reducing reliance on third-party LLM providers while improving cost and speed.

Before Luminance, I researched context effects in dialogue models at the University of Aberdeen and The Context Lab. My first-author paper on local context repetition in LLM utterance production and comprehension appeared at CoNLL and was presented at EMNLP.

My route into machine learning started with engineering. I built robots and sensors as a teenager, published my first scientific work before university, studied electrical engineering in Budapest, worked on national-scale telephony systems at Ericsson, and later completed a first-class honours BSc in Computing Science in Aberdeen. I’ve worked across applied ML, healthcare, defence, drone systems, automation, and production software.