YEJOONG KWON

Airflow Migration to Kubernetes
Apr 2025 – Jun 2025

• Migrated the existing EC2 + Docker-based Airflow infrastructure to a Kubernetes-based Helm Chart setup
• Refactored the structure to enable individual Docker image builds and deployments per DAG, accommodating different dependency versions
• Separated compute resources by DAG execution unit to prevent conflicts in package contents and versions between DAGs
• Achieved stable operations for workloads with large data volume fluctuations, such as CDC deduplication tasks

Migration of Analytical Data Mart Transformations to DBT
Apr 2025 – Jun 2025

• Standardized fragmented transformation workflows of analytical data marts previously managed via Databricks UI to a DBT + Airflow-based pipeline
• Modularized and refactored transformation logic for over 100 silver/gold mart tables to enable reusability and consistency
• Introduced Git-based version control to prevent workflow loss due to human errors
• Ensured idempotency in batch data mart generation, eliminating the need for full workflow re-runs in case of data loss or failure, reducing operational time and cost

Data Warehouse (DW) Operations & Maintenance
Jan 2024 – Dec 2024

• Built a data consistency dashboard to validate alignment between operational and analytical databases
• Monitored and troubleshot data migration pipelines (Airbyte–Kafka, Airflow)
• Led internal training sessions and provided ongoing support on querying and utilizing the EDW
• Created and maintained shared view tables and continuously updated data documentation
• Contributed to breaking down data silos and restoring company-wide trust in data quality

Activation of DW Usage through External Data Integration
Apr 2024

• Designed pipelines to ingest external datasets—such as weather data, administrative boundaries (GeoJSON), depot address information, and local place data—into the data warehouse (DW)
• Established a foundation for advanced data analysis, leading to active utilization across various departments by enhancing the effectiveness of marketing and strategic planning through integration with internal operational data

Building an Analytics Environment for Customer App Event Logs
Oct 2024 – Dec 2024

• Previously, customer app event logs were stored in JSON format but were not utilized for analysis
• There was ongoing demand from related departments (CRM, Planning) to leverage this data.
• Built a near real-time data warehouse environment by converting and storing the logs in Iceberg table format, enabling analysis through Trino

Reducing Database Costs through Migration of Vehicle Tracking Log Data
May 2024

• Addressed the need to reduce costs for vehicle tracking logs, which were being generated and stored at over 3 million records per day.
• Migrated data from the existing database to cost-effective S3 storage with compression, reducing cost of storage usage by 88%.
• Adopted the Iceberg table format to simplify schema management and improve query performance.
• The migrated data is now actively used by the vehicle tracking development team.