Six years building product analytics for a blockchain platform, within a decade across data and engineering. Now finishing an MSc Computer Science with a dissertation on fraud detection using dbt, BigQuery, XGBoost and SHAP.
Ten years across product, growth, and engineering.
I work at the intersection of analytics, product, and engineering. Six years inside a high-growth blockchain platform taught me to turn messy behavioural data into decisions leadership can act on — the same problem at the core of UK fintech, iGaming, and product analytics teams.
I design KPI frameworks that survive scale, build behavioural segmentation that drives measurable engagement, and own the data quality standards that leadership decisions depend on. The Experience and Work sections below have the details.
Shipped products, MSc research, and side explorations.
Product Owner and backend developer in a seven-person Scrum team. Designed system architecture, led backend setup with FastAPI and PostgreSQL, built core commitment and group logic. Delivered 67 automated tests (100% pass rate) and an 8.8/10 UAT satisfaction score.
End-to-end pipeline on the IEEE-CIS dataset (~590K transactions): BigQuery → dbt feature layers (staging / intermediate / mart) → Optuna-tuned XGBoost → SHAP explainability → Power BI dashboard, orchestrated with Airflow and containerised with Docker. Tests whether a structured dbt transformation layer measurably improves model performance over raw features, with an LLM layer summarising SHAP into plain-language risk explanations.
Interactive dashboards visualising Indian election and budget data. Raw data scraped and structured from ECI results portals, data.gov.in, and official government press notes. Built as single deployable HTML files — no server, no backend, no dependencies.
A decade spanning systems engineering to analytics leadership.
MSc Computer Science, with a dissertation on fraud detection and machine learning.
Fraud Detection Pipeline Using Analytics Engineering & Machine Learning
Tests whether disciplined analytics-engineering — structured dbt feature layers on BigQuery — measurably improves fraud-detection performance over raw inputs, evaluated with XGBoost and SHAP on the IEEE-CIS dataset. Supervised by Dr Bernie Tiddeman. The full build is featured in the Work section above.
End-to-end data stack from raw SQL to ML explainability.
I'd welcome a conversation about analytics roles — or anything interesting where data meets product.
References available on request · Typical response time within 24 hours