Data Analyst | SQL · Power BI · Python | Fraud & Financial Analytics | Monash MDS
Master of Data Science, Monash University (Dec 2025). Melbourne, Australia. Open to data analyst and data engineering roles.
My current project asks one question: at what risk-score threshold does manual fraud review pay off in money? (fintech-fraud-triage, on ~688k transactions of Revolut-style open data.) Before the MDS I spent 13 months running month-end settlement cycles at S&I.
- ✅ kinvest-product-analytics: shipped Aug 2026 — SQL-first product analytics (funnel · cohort retention · LTV · RFM) on a self-designed synthetic event log
- 🚧 fintech-fraud-triage: Stage 01 (EDA + user feature table) complete; leakage-safe feature set in progress
- ⏳ PL-300 (Microsoft Power BI Data Analyst): preparing since February 2026 on the official Microsoft Learn track
- ⏳ Dynamics 365 Finance fundamentals: Microsoft Learn path, in progress
| Project | Stack | Highlights |
|---|---|---|
| Fintech Fraud Triage 🚧 | Python | Cost-sensitive review-threshold framework on Revolut-style open data (StrataScratch): 9,944 users, 3.00% fraud prevalence, 688,651 transactions. Stage 01 done — EDA with integrity gates, a contract-tested 117→9 merchant-category mapping, a 39-column user feature table. Now building the leakage-safe feature set; then PR-AUC vs the 3% prevalence floor and a net-savings threshold sweep |
| K-Invest Product Analytics | SQL (PostgreSQL), Python | Self-designed mobile-brokerage event taxonomy (GA4/Amplitude-export-compatible) → benchmark-anchored synthetic event log (15,000 users, 90 days, ~384k events, seed=42) → SQL-first funnel, weekly cohort retention, LTV curves, RFM segmentation; D1/D30 calibrated against public benchmark bands. Openly synthetic: the deliverable is the methodology |
| Melbourne Transport Accessibility Analysis | SQL, PostGIS, Tableau | 360 suburbs, 59,000+ stops; route coverage R²=0.88 vs stop density R²=0.57. Validation rules documented in the repo |
| Marketing Campaign Causal Evaluation | Python, Power BI | A/B testing on 41,188 records. Logistic regression; OR 1.45 [1.35–1.55], p<0.001 |
| Digital Nomad Destination Dashboard | R Shiny, Leaflet, ggplot2 | 40+ countries; multi-index scoring; K-means clustering |
| Job Engine Web | Next.js, TypeScript, Supabase, Claude API | The pipeline I run my own job search on: automated collection (GitHub Actions cron), Claude fit-ranking, per-role resume tailoring. Live on Vercel |
| Status | Credential / learning path |
|---|---|
| ⏳ In progress | PL-300: Microsoft Power BI Data Analyst, preparing since February 2026 (official Microsoft Learn track) |
| ⏳ In progress | Learn the fundamentals of Microsoft Dynamics 365 Finance, Microsoft Learn learning path |
| ✅ Completed | Prepare data for analysis with Power BI, Microsoft Learn learning path (March 2026) |
| ✅ Completed | Get started with Microsoft data analytics, Microsoft Learn learning path (February 2026) |
Languages: Python (pandas, NumPy, PySpark), R (dplyr, ggplot2, Shiny), SQL (PostgreSQL, PostGIS) Analytics & BI: Power BI, Tableau, Excel (advanced) Data Engineering: ETL pipelines, PySpark, Apache Kafka, Docker, FastAPI Tools: Git/GitHub, Jupyter, VS Code
FIT5149 Stock Volatility Prediction: 2nd place on the Kaggle private leaderboard (Monash University, Sep 2024)
Master of Data Science, Monash University, Dec 2025. Distinctions in Database Design, Data Visualisation, Data Wrangling, Foundations of Data Science, and Applied Mathematics for ML.
BA (Global Business & Technology + Polish), Hankuk University of Foreign Studies, 2016–2021.