← All projects
RWA Calculation Engine
Built from scratch the platform that calculates Risk-Weighted Assets for a tier-1 bank's Risk department — powering monthly IRB regulatory reporting to the Central Bank.
Senior Data Engineer · Raiffeisen Bank
PySparkAirflowApache IgniteSpring Cloud Data FlowPythonSQL
A from-zero product for the Risk department: compute Risk-Weighted Assets (RWA) across retail and non-retail portfolios on the IRB approach, accurately and on a strict monthly cadence — the output goes to the Central Bank of Russia, so correctness is non-negotiable.
What I did
- Designed the DWH architecture and data products behind the calculation.
- Built ELT/ETL pipelines (PySpark / Python / SQL) integrating many internal sources, orchestrated with Airflow and Spring Cloud Data Flow.
- Migrated Spark 2.4 → 3.2 and tuned the JVM, improving pipeline performance by ~50%; further sped up Ignite with targeted Java options.
- Created a PySpark unit-testing framework and data-quality checks — regression safety for a regulated calculation.
- Delivered monthly RWA reporting (IRB) to the Central Bank, plus custom integrations with internal APIs.
Impact
A dependable, well-tested regulatory engine that ran month after month — fast enough to fit the reporting window and trustworthy enough for a central-bank submission.