xavier113

Built and maintained real-time streaming pipelines to process Hoyoverse payment events, including payment orders, refund orders, and application logs. Designed and optimized offline data models through dimensional modeling, improving data warehouse design for payment, access control, risk control, revenue, reconciliation, and refunds. Integrated data from multiple overseas SDK services and payment gateways, ensuring schema standardization, data validation, and robust ingestion workflows. Improved query performance and reduced compute/storage costs through model redesign, table optimization, and job scheduling improvements. Established monitoring and DQC for data freshness, pipeline stability, and reconciliation consistency.

Supervised the organization of tracking data, led data platform construction, and optimized pipelines, resulting in enhanced tracking data life-cycle management. Designed and built an ETL pipeline capable of processing over 200 billion daily events while concurrently supporting millions of users. Implemented a low-latency(ms) ETL pipeline for rapid event delivery to critical teams, enabling real-time decision support. Established Shopee app devices and implemented a data pipeline to capture reinstall data, eliminating the need for Appsflyer. Led the development of a cutting-edge real-time/offline tracking data warehouse using dimensional modeling techniques, enabling enhanced data analysis capabilities. Engineered a robust data quality monitoring platform, integrating real-time monitoring, offline monitoring, and predictive analytics by Machine Learning to ensure data validity and provide multi-dimensional risk alerts. Established a user-centric data dashboard displaying crucial metrics, including UV, PV, and session-related indicators, facilitating data-driven decision-making. Engineered a configuration center to streamline data cleaning and routing, resulting in resource allocation optimization and enhanced data processing efficiency.

Led backend design and development efforts, facilitating data persistence through Kettle. Innovated an integrated SDK with Impala as the entry point, abstracting underlying storage complexities and enabling data manipulation and retrieval.