Open to opportunities

Kevin Chen

@kevinchen1

Message

Senior Data Engineer bridging large-scale data platforms and applied AI for reliable production systems.

United States

Message

What I'm looking for

I seek senior data platform or applied-AI engineering roles where I can build scalable, reliable pipelines, mentor teams, and drive measurable product and compliance outcomes.

I am a Senior Data Engineer with 9+ years designing and scaling high-performance data platforms, multimedia pipelines, and secure analytics systems for global consumer products and enterprise marketing. I focus on reliability, compliance, and cross-team collaboration to deliver measurable impact.

At TikTok I engineered Kafka-to-Flink pipelines, large-scale Spark and Delta Lake ETL, and multimodal feature flows that supported billions of daily events and improved processing and experimentation latency. I standardized online/offline feature flows across Redis, Cassandra, and ByteHouse while reducing feature drift and storage costs.

Previously, I built consumer identity and linkage systems (CLINK™/DLINK™) that achieved 96%+ match accuracy, integrated orchestration and monitoring with Dagster and Prometheus, and enabled 1,000+ predictive models annually. I mentor engineers, contribute to open-source efforts, and prioritize explainability and compliance in production systems.

Experience

Work history, roles, and key accomplishments

Current

Senior Data Engineer

Current

TikTok

Mar 2022 - Present (3 years 10 months)

Engineered Kafka-to-Flink and Spark+Delta Lake pipelines to process 10B+ daily events, improving historical feature processing time by 35% and reducing feature drift 40% while enabling sub-second online feature freshness for continuous model updates.

Python Kafka Flink Spark Delta Lake Clickhouse Redis Cassandra Datadog

Senior Developer

DataLab USA

Feb 2015 - Dec 2021 (6 years 10 months)

Built and scaled CLINK™/DLINK™ consumer identity pipelines using C#, Python, and Databricks, achieving 96%+ match accuracy, reducing identity conflicts 35%, and delivering features for 1,000+ predictive models annually while ensuring FCRA compliance.