Data platform modernization: open table format serving

• Migrated the PostgreSQL-centered API serving layer to Delta Lake on Databricks to meet the varied data consumption needs of institutional clients, including US brokerages
• Exposed an Iceberg catalog through Delta UniForm so clients can connect Snowflake and BigQuery directly, opening a data sales channel beyond the API

Analytics pipeline standardization and improvement

• Managed and optimized the data pipelines of the BigQuery-based analytics platform
• Standardized transformation pipelines with dbt and moved large-scale computation to distributed Spark processing, shortening data latency

XRP Ledger data product

• Built pipelines that collect XRP Ledger and market data into PostgreSQL and BigQuery
• Secured a competitive data edge with real-time updates and transaction data decoding

Workflow orchestration and infrastructure

• Built and operated HA Airflow managed with ArgoCD, Kustomize, and Helm, securing pipeline stability and operational efficiency
• Moved EC2 services to EKS and introduced Karpenter autoscaling; provisioned batch workloads quickly and flexibly with SPOT instances and Kubernetes restart logic

Integrated pipeline for high-volume user logs

• Developed and operated pipelines that integrate and preprocess high-volume user logs in real time across KakaoTalk, Gift, advertising, Daum News, and Daum Story
• Developed and maintained recommendation systems for multiple services, improving user engagement and experience

Led the adoption of Airflow 2.0

• Led the adoption of Apache Airflow 2.0, improving the scalability and operational efficiency of task orchestration (presented at if(kakao))

Search keyword ad recommendation pipeline

• Built a data pipeline that predicts pCTR from Daum Search query-consumption history and recommends keyword ads
• Reached responses within 50 ms with vector DB (ANN) serving

• Designed and ran a delta-hedged market-making strategy that combines Uniswap LP positions with futures positions
• Implemented synthetic strategies on futures and options, and automated position management and risk monitoring
• Built pipelines that collect and analyze on-chain and off-chain market data, used for signal generation and strategy backtesting