Most of what I know came from building the thing and watching it fail.
Data Engineer London
Summer Data Analyst, JTP Shenzhen
OA and Data Software Development Shenzhen
MSc thesis · KCL
MSc thesis. A reproducible backtesting framework over five Uniswap v3 pools showing that walk-forward-selected liquidity-provision configurations don't beat a never-rebalanced position, and that overfitting diagnostics change sign with transaction costs.
Team research project (4) · KCL
Discrete-time agent-based model of heterogeneous BTC traders reacting to Polymarket belief signals. Stronger herding pushes order imbalance to extremes and flips price dynamics from noisy to persistent one-sided regimes.
Published research · University of Nottingham
Published BSc research. A multi-agent RL trading system with a DQN agent learning alongside diverse rule-based agents. It made 12.23% money-weighted annual return and 15.9% cumulative return on unseen S&P 500 data with low trade volatility.
Industry project · Capgemini
End-to-end credit-evaluation model built at Capgemini (Shenzhen): preprocessing, binning, feature selection, training and testing, at 81% classification accuracy.
Mentor-led research · supervised by Prof. Antoine Jacquier, Imperial College
Mentor-led research with Prof. Antoine Jacquier (Imperial). LSTM models forecasting daily US-equity returns; owned the pipeline end to end from data cleaning and feature construction through walk-forward evaluation.
Software project sponsored by BJSS, Nottingham
Led prototype development as co-leader, designing against the sponsor’s requirements in Figma and TypeScript, and ran code and milestones as the team’s Git admin on Jira and GitLab.
Software projects, Nottingham
Reversi with real-time multiplayer and an AI computer player, then a second game on an MVC structure in JavaFX with a customisable UI, a scoring algorithm and a database-backed live leaderboard.
Software project, Nottingham
A movie database designed from scratch with one-to-many relationships, full add, delete and manipulate operations and a search algorithm, with its web front end built against it.
Group project, Nottingham
Worked in Agile against stakeholder needs, leading prototype and UI design, developing and testing in Java with JUnit, and coordinating the group on GitLab.
Independent project, adopted by Haileybury Middle School
An interactive, to-scale galaxy simulator built alone, learning C# from zero and expressing the physics and orbital maths in code. Published on Unity Play and adopted into Haileybury Middle School’s physics curriculum, with an 83% student satisfaction rate.
Research project led by Professor Mark, MIT
In a team of four, collected 600+ data points from the transport maps of major Chinese cities and used SVM, logistic regression and neural networks to train route-planning models and build a digital map of Shanghai. Scored 90%.
MSc Computational Finance
Quantitative Methods in Finance · High-Frequency Finance · Statistics in Finance
BSc Computer Science, Upper Second-Class Honours (2:1)
Machine Learning · Mathematics for CS · Linear Discrete Optimisation · AI Methods · Front and back-end development · Image processing
A-Levels and IGCSEs
A-levels: Maths (A*), Physics (A), Computer Science (A), Chinese (A*)
11 GCSEs, six grade 9s including Maths and Physics. IELTS 7 (2020).
Represented the course to the department and to prospective students.
The society was new in my year. I taught myself C# and Unity and then taught the society, running sessions on things nobody was covering in class including Python, an introduction to the web and general computing, open to every year group from 9 to 12.
Also a lower school mentor.
Olympiad preparation by working problems together; organised an inter-school maths competition and maths magic classes.
Four years in the team, two of them as captain, and we did not lose a game across those two seasons. Haileybury is 40% international, so leading it was as much about communication as about basketball.
I still play. The songs are over on the Life pages.