London, England, United Kingdom
Pushing the boundaries of Generative AI and finding novel applications in a rapidly developing field. With a background that includes Neuroscience, Biology, LLMs, I’m highly interested in applying computational theory and practical developments across disciplines.
Innovating LLM-based agentic frameworks and interpreting foundation model form and function for Nature data. - Leading data science strategy for a sustainable FinTech company, steering a team using the latest advances in a rapidly evolving AI landscape. - Combining state-of-the-art LLM models and agentic frameworks for AI driven data extraction and verification. - Optimising financial portfolios to adjust for nature-related risk. - Training multi-modal foundation models combining geospatial data with natural language. (huggingface, pytorch)
- Training and fine tuning large language models (GPT, BERT, T5, LangChain, with LoRA models) using quantisation and multi-GPU setups in Google Cloud. - Shaping the GenAI strategy for the global data science team -Rapid development of front-end interfaces to bring the developed science to life. - Developing optimisation packages using OR Tools and non-linear algorithms.
Generating data science solutions for retail business problems. Using python and Gitlab to create productionisable packages and collaborate with users and clients to optimise existing products. - Designing and training LLM on a custom dataset using PyTorch and Huggingface - Optimising distributed GPU processing - Developing and productionising machine learning, network analysis, and NLP packages using Python, Pyspark, SQL, and Git - Awarded the annual global data science award 2023 for innovation for a project on training transformer models from scratch using retail data.
Science to Data Science This is an intensive 5 week data science workshop designed to assist academic researchers transfer their skills to data science. During the programme I worked in a group of five on a project with Pivigo. The aim was to analyse their data with a variety of machine learning and natural language processing techniques to extract non-traditional metrics from company reports using natural language processing (NLP).
Designing and leading a major neuroimaging study and handling responsibilities beyond my training level. Processed time series data and applied machine learning to characterise cognitive and visual processing in neural data.
Using questionnaire data and memory performance to investigate how mood and ageing influences behaviour. - Published research in peer reviewed journal (Published as Mok, Hajonides et al, 2019 Emotion) - Selected as a national top-five undergraduate thesis in biological sciences