Dariusz Dyrga

Led a full-system cutover migration of cost, sales, and client data from a legacy SQL system to BigQuery, translating complex business logic into the target schema. Built Cloud Run Java ETL jobs, BigQuery views, and Informatica IICS workflows to orchestrate and support the migration workflow with business stakeholders.

Architected and delivered an end-to-end streaming data lake for a financial institution processing hundreds of Kafka topics with under 1 TB/month of net new data across Bronze, Silver, and Gold layers. Implemented Kafka and AWS Glue schema governance, built an Airflow DagFactory orchestration pattern for hundreds of pipelines, and defined repeatable infrastructure using Terraform and Terragrunt.

Led end-to-end design and delivery of a custom AWS data platform integrating approximately 45–50 data sources to replace FiveTran, enabling lower-cost ETL and a foundation for future ML workloads. Built distributed PySpark ETL on AWS Glue, migrated GA3/GA4 into Redshift with schema redesign and nesting flattening, and orchestrated workflows using Glue Workflows and Step Functions.

Developed Java integration code connecting internal systems to Databricks for batch data cataloguing from BigQuery. Optimized BigQuery SQL queries and built/maintained Databricks ETL notebooks using GitLab for CI/CD.

Designed and implemented GCP API and legacy-system ingestion pipelines into BigQuery for executive-facing Looker dashboards covering cost reporting, resource metrics, and security compliance. Resolved BigQuery performance bottlenecks by partitioning/clustering and splitting a large table, reducing a key query from 30 minutes to 20 seconds and cutting monthly BigQuery costs from $6,000+ to under $1

Supported training of a Two-Towers AI recommendation model by building SQL-driven feature tables and orchestrating data movement between BigQuery and BigTable. Implemented Kubeflow pipelines with logging, staging, and validation steps, and added data-quality checks with unit tests and GitLab CI/CD automation.

Implemented FiveTran-based ingestion combined with BigQuery to address growing multi-source data volumes and prepare datasets for analytics consumption. Refactored an Airflow DAG management approach by introducing a DagFactory pattern with YAML configurations and built pipelines migrating data from BigQuery to BigTable.

Implemented data streaming pipelines using Amazon Kinesis and maintained a Scala-based ETL application. Strengthened AWS fundamentals by completing the AWS Certified Cloud Practitioner exam.

MSc in Computer Science at Jagiellonian University with a thesis on transfer learning applied to coronary angiography image analysis.

BSc in Computer Science at Jagiellonian University with a thesis on a patient findings data platform as a REST API (Java/Spring) with a React/MaterialUI frontend.