MSc Data & Computational Science (UCD) | Data Analyst | Applied Machine Learning
Data science graduate working on real-world datasets, turning raw data into models and insights that support decisions. Currently job-hunting in Dublin for data analyst / ML roles.
| Category | Tools |
|---|---|
| Programming | Python, R, SQL |
| Data Analysis | Pandas, NumPy, Exploratory Data Analysis |
| Machine Learning | Regression, Classification, Model Evaluation |
| Visualization | Matplotlib, Seaborn |
| Other | Git, Jupyter Notebook |
Fraud Detection in Financial Transactions
With Mani Jose, UCD School of Mathematics and Statistics. Real-time fraud detection on the BankSim dataset (594,643 transactions, ~1.2% fraud rate). Engineered 39 behavioral/temporal features, benchmarked 4 modeling approaches, and deployed the best LightGBM ensemble via a FastAPI scoring endpoint + Streamlit monitoring dashboard.
- AUROC 0.9993 | AUPRC 0.9578 | Inference latency < 10ms
Compared 18 years of official CSO rent data against ~500 live Daft.ie listings (scraped with Playwright) to quantify what new movers actually pay vs. the published average — framed as a relocation-budget tool for HR teams. SQL (window functions, CTEs) + Power BI dashboard.
- €2,139 avg rent (2025) | +112% vs. 2011 low | ~21% avg "newcomer premium" over official rates
Customer Shopping Behavior Analysis
End-to-end EDA, SQL, and Power BI dashboard on 3,900 transactions across 25 products and 50 locations. Found male customers drive 68% of revenue ($157,890 vs $75,191) and isolated 839 high-value discount users (21.5% of discount users) as a targeted upsell segment.
- Tools: Python (pandas), PostgreSQL, Power BI Zepto SQL Case Study
SQL-only exploration of Zepto's product dataset — cleaning, null handling, and 8 analytical queries covering revenue estimates, discount strategy, and inventory-weight distribution.
- Tools: PostgreSQL, Excel
- 🔭 Open to Data Analyst / Junior ML roles (Dublin)
- 📚 Learning: [REPLACE — e.g. specific cloud/ML cert in progress]
