At Bank of Georgia, I co-led the migration of 40+ legacy data projects from Hadoop to Spark and MinIO on Kubernetes, reducing run times by 30–40% and storage by 20–25%. I build batch and near-real-time pipelines and maintain an on-prem Iceberg lakehouse of about 60 TB.
I design analytical and feature data models across Oracle, SQL Server, and PostgreSQL, and build data quality checks and freshness alerts. My work also includes a daily pipeline moving mBank events from BigQuery into the lakehouse and monitoring calculations that went from taking 2–4 hours to 5–15 minutes.
I build and monitor risk assessment APIs and model-serving services on Kubernetes, as well as platform tools for data scientists. I also contributed to an internal platform used by about 90 colleagues, including a text-to-SQL chatbot with RAG, and share AI tools and practices with engineers as an AI Ambassador.

