Projects

Selected projects across Applied Data Science and AI Engineering.

Each project includes links to code, live demos/APIs where available, and measurable results.

Applied Data Science & ML Systems

Diabetes Readmission Prediction (MLOps Production System)

End-to-end MLOps system predicting 30-day hospital readmission risk at discharge with live API + Streamlit dashboard, monitoring, explainability, and ROI analysis.

Reduced preventable readmissions with projected £7.9M annual savings

PythonPandasNumPyscikit-learnLightGBM+12

GitHub ↗ Demo APIDetails

◎ Demo Soon

E2E Time Series Forecasting & Analytics System

Production-grade time series forecasting platform with leakage-safe ETL pipelines, recursive multi-step prediction, rolling-origin backtesting, and FastAPI serving — deployed on GCP Cloud Run.

Production forecasting over 811K+ rows across 1,115 stores

PythonLightGBMPandasFastAPIPydantic v2+9

GitHubDetails

○ No Demo

Contract Review & Risk Analysis System (CAUD / CUAD)

AI-powered contract analysis and risk scoring system with live Streamlit dashboard + Cloud Run FastAPI API, batch processing, CI/CD, and monitoring.

Automated contract risk detection with 97% contract-type accuracy

PythonFastAPIStreamlitscikit-learnTF-IDF+9

GitHub ↗ Demo APIDetails

○ No Demo

MSc Thesis: Training Set Optimisation for Linear B-Cell Epitope Prediction

MSc thesis optimizing training datasets and feature selection for Linear B-Cell Epitope prediction, benchmarking XGBoost vs neural networks with strong AUC/F1/MCC performance.

Achieved 99.4% AUC-ROC with optimized training sets for vaccine development

PythonXGBoostTensorFlow/KerasAzure MLBoruta+5

GitHub ThesisDetails

RAG & Agentic Systems

● Live

Legal Q&A RAG Chatbot

Production-ready RAG chatbot for UK legal queries, processing 131,253+ chunks with sub-3s latency, hybrid retrieval (BM25 + FAISS + RRF), enterprise auth, and guardrails.

Enabled sub-3s legal search over 130k+ documents

PythonFastAPIDockerPostgreSQLAlembic+12

GitHubDetails

○ No Demo

ADCO: Autonomous Data & Compliance Officer (Multi-Agent System)

Production-grade multi-agent system for automated GDPR/HIPAA/CCPA compliance audits using Google ADK + Gemini + Presidio + ChromaDB with rigorous evaluation.

Automated GDPR/HIPAA/CCPA compliance audits with 85%+ precision

Python 3.11Google ADKVertex AI (Gemini)ChromaDBMicrosoft Presidio+7

GitHubDetails