London, UK · GitHub · LinkedIn · Full right to work in UK & EU
Download CV (PDF)Final-year BSc Computer Science student at the University of Surrey (predicted First Class), graduating July 2026 and available from September. A strong Python developer with applied experience spanning AI/ML, full-stack development and Linux infrastructure — from fine-tuning and evaluating LLMs to building and self-hosting containerised services. Comfortable across the engineering lifecycle: Git, automated testing, code review, CI/CD and Agile.
Built a three-stage co-evolutionary RL pipeline (40M timesteps) training cooperative agents to stay effective alongside imperfect or adversarial partners in Overcooked-AI. Evaluated across three seeds with a compute-matched ablation study, quantifying a 2–3× improvement over the self-play baseline and isolating which stages drove the gains.
Implemented NSGA-II for multi-objective hyperparameter optimisation of an LLM, balancing validation accuracy against trainable parameter count. Analysed the resulting Pareto front to find a deployment configuration that cut parameters by 55.5% with matched accuracy.
Prototype assistive-vision system that processes camera or webcam frames through a multi-stage pipeline: detects accessibility-relevant objects with a fine-tuned YOLO11 model, estimates proximity with Depth Anything V2, scores risk, and outputs text-to-speech alerts plus annotated logs. Configurable via YAML, with evaluation scripts benchmarking model sizes on COCO and custom obstacle datasets.
Local RAG tool that melts and pre-aggregates large public London crime datasets with pandas to make lookups near-constant-time. Cut local-LLM hallucinations during conversational retrieval through regex-based entity extraction and precise context prompting.
Led a 7-person Agile team to design, build and deploy a containerised e-learning platform, standardising Git workflows, group code reviews and automated testing. Built the React frontend with secure JWT auth and RESTful endpoints, and accessible high-contrast UIs tailored to an elderly demographic.
Rebuilt one working machine from two scrap HP laptops, installed Linux Mint, and run it as a headless, Docker-based home server over SSH. Full setup documented as a rebuild/recovery runbook, including a layer-by-layer network debugging write-up.
Access to HE Diploma in Computing: Distinction (45 credits at Distinction,
15 at Pass).
A-Levels: Mathematics (A), Physics (B), Economics (C).
Delivered high-standard hospitality for clients including KPMG and Microsoft in fast-paced environments, with consistent reliability under pressure.
Organised 500+ exam papers into a structured folder hierarchy, improving access and demonstrating strong organisation.
Pulled and analysed financial metrics across 10+ companies' annual reports to support investment decisions; early exposure to a regulated finance environment.
Provided high-quality customer service in a busy café setting, demonstrating reliability and teamwork.