I'm a data scientist and AI engineer specialising in machine learning for financial services and LLM-powered applications. I build Python-based systems that help fintech companies, lenders, and businesses turn raw data into decisions — and increasingly, into intelligent, conversational tools.
Financial ML
- Credit risk modelling — probability of default, scorecard development, risk tiering
- Fraud detection — anomaly detection, transaction classification, imbalanced learning
- Customer churn prediction — behavioural analytics, survival analysis, retention strategy
LLM & AI Engineering
- RAG systems — document Q&A, knowledge-base chatbots, retrieval pipelines (LangChain, vector databases)
- AI agents — tool-using assistants that reason, retrieve, and act on real data
- LLM application development — chains, structured output, conversational memory
Foundations
- Python data pipelines — pandas, scikit-learn, XGBoost, feature engineering, model evaluation
| Project | Description | Key result |
|---|---|---|
| Credit Risk Scorecard | End-to-end default prediction on 150k borrower records | AUC-ROC ~0.82 |
| Fraud Detection System | Transaction fraud classifier with imbalance handling | Coming soon |
| Fintech Agent | AI agent that answers financial questions using live market data and calculation tools | Chains multiple tools autonomously e.g. fetches live price, then computes ROI, without manual orchestration |
| Document Q&A Assistant | RAG-powered chatbot that answers questions from a document knowledge base | Deployed app with grounded, source-cited answers. Correctly declines unanswerable questions |
languages = ["Python", "SQL", "R"]
ml = ["scikit-learn", "XGBoost", "pandas", "numpy"]
llm = ["LangChain", "LangGraph", "Gemini", "Groq", "RAG pipelines", "agents", "tool-calling"]
backend = ["FastAPI", "Pydantic", "SQLAlchemy"]
frontend = ["React", "Vite"]
viz = ["matplotlib", "seaborn", "Power BI"]
tools = ["Jupyter", "Git", "VS Code", "Excel", "Power BI", "Vercel"]- Building a fintech ML portfolio — credit risk, fraud, churn
- Building LLM/RAG applications — document Q&A, conversational agents
- Going deeper on model evaluation — ROC-AUC, SHAP, threshold optimisation
- Available for freelance data science and AI engineering projects on Upwork
If you're working on a financial data problem or need an AI/LLM-powered tool, let's talk.
