MSc Data Science · University of Exeter
Data Engineer and Software Engineer specialzing in scalable financial systems, Python, and C++.
A matching engine built in C++20 replicating core exchange infrastructure. Implements price-time priority matching, O(1) order cancellation via hash-indexed lookup, and fixed-point integer price representation to eliminate floating-point map key errors. Includes VWAP tracking, nanosecond-resolution latency measurement, and test suite covering limit orders, market orders, partial fills, and cancellation.
GitHub →
MSc dissertation investigating TabNet and FT-Transformer for daily equity trading signal generation across ~300 S&P 500 constituents (2015–2025). Evaluated via a daily-rebalanced long-short strategy benchmarked against SPY, with an interactive backtesting dashboard.
Interactive RAG application that indexes my CV and projects for semantic retrieval. Deployed live on Streamlit Cloud - try asking it about any of my experience.
Benchmarked MLP, LSTM, and Transformer architectures on financial news sentiment classification. Transformer models delivered the highest performance with 90.3% Accuracy, 0.876 F1 Score, and 0.969 ROC-AUC, outperforming RNN baselines across all metrics.
Engineered a normalised (3NF) relational database on Azure SQL to analyse UK agricultural sustainability. Built automated Python ingestion pipelines and a Node.js REST API layer.
Advanced LLM applications coursework comparing full fine-tuning vs. LoRA on DistilBERT, and implementing RAG for question answering. Achieved 89.8% accuracy with full fine-tuning and 85.7% with LoRA (49% faster, 99% fewer parameters). RAG system improved answer quality by 8.5× using retrieved context.
GitHub →
End-to-end credit risk pipeline on real-world loan data. Compared Logistic Regression, Random Forest, and XGBoost; addressed multicollinearity and class imbalance. XGBoost achieved the best results with 92.8% Accuracy, 0.944 AUC, and 0.816 F1-score.
Implemented a full MLP in NumPy - forward pass, backpropagation, weight updates - achieving ~89% accuracy on Fashion MNIST, on par with a reference PyTorch implementation.
GitHub →
Identified top clinical predictors of diabetes diagnosis and benchmarked classification models. Logistic Regression yielded the highest accuracy at 77.3%, while SVM achieved the highest AUC at 0.850. Includes feature importance analysis and model interpretability.
Semi-automated reporting pipeline built during internship at Ekimetrics: raw data ingestion and cleaning through to interactive Power BI dashboards for commercial KPI tracking.
Proprietary - not publicly accessible
Third-year dissertation evaluating machine learning applications in T1D: glucose forecasting models, diagnostic tools, and closed-loop insulin delivery systems. Highlights include XGBoost achieving 0.99 AUC for hypoglycaemia prediction and dual-hormone systems reaching 93.1% Time in Range (TIR).
Read report →
Evaluated the Diabeloop DBLG1 AI-driven diabetes management system. Analyzed algorithmic bias, transparency, and fairness through utilitarian and principlism frameworks, proposing collaborative governance strategies.
Read report →
Streamlit web app for processing student data exports using a Bronze → Silver → Gold medallion architecture. Cleans and validates records, then splits into four exports for Power BI, membership, and survey platforms. Hosted on Azure App Service.
Proprietary - not publicly accessible
12-module certification covering Python, SQL, data visualisation, machine learning, and applied data science. Verified by IBM via Coursera.
View certificate →Email is the fastest way to reach me.