Aspiring data engineer. I build ingestion systems and data products from messy public / market sources — mostly Sri Lanka.
Looking for Data Engineering roles where I can own pipelines end-to-end: extract → model → serve → keep them honest.
Portfolio · LinkedIn · Email · Ardeno Studio
Only three projects. Everything else is noise.
1. PropertyLK · live
Sri Lanka real-estate price intelligence
Scrapes listings, cleans and geocodes them, aggregates market signals, and serves deal scoring / heatmaps.
| Proves | Messy multi-source ingestion, cleaning, geocoding, scheduled pipelines, serving layer |
| Stack | Python · FastAPI · PostgreSQL · Playwright · React |
| Why it matters | Real local market data with a usable product on top — not a tutorial dataset |
Fuel price tracker + open API for Sri Lanka
Scheduled scrapes for CPC / LIOC prices, history, trip cost calculator, email alerts, public API.
| Proves | Recurring ingestion, source fragility, API design, alerting, CI around scrapers |
| Stack | TypeScript · FastAPI · Fly.io · Vercel |
| Why it matters | Narrow scope, ships continuously, and survives real source-website breakage |
3. Vega
Streaming lakehouse playground
Kafka → Kafka Connect → Flink → Iceberg on Azure AKS, with connectors, Terraform, and tests.
| Proves | Streaming / lakehouse mechanics I am actively learning in depth |
| Stack | Java 21 · Kafka · Flink · Iceberg · AKS · Terraform · dbt |
| Honest note | Strongest infra project, still a portfolio system — not claiming production SLAs |
- cse-api-docs — Unofficial Colombo Stock Exchange API map, live-probed weekly (docs)
- DataFlow — Kafka → Spark → Delta → dbt → Airflow medallion pipeline (Olist)
- Koel — Telegram alerts for CSE price moves (live)
I also maintain a set of reverse-engineered Sri Lanka API notes (banks, utilities, markets). Useful for local data work; not the hiring pitch. Browse.
Python · SQL · Java · TypeScript
Kafka · Flink · Spark · Iceberg / Delta · dbt · Airflow
PostgreSQL · Docker · Terraform · GitHub Actions · FastAPI
- Prefer one honest pipeline over ten demos
- Document source provenance, freshness, and known gaps
- Care about retries, idempotency, and tests — still leveling these up on every project
- Bias toward local market data where public sources are messy and under-documented
Open to Data Engineering roles / internships. Happy to walk through architecture, failure modes, and what I would harden next on any of the projects above.