Lu Zheng

Kafka exactly-once deletion platform processing 10M+ req/month with 99.95% SLA. Built Flink stream processing, Airflow pipelines, and Kubernetes multi-region EKS infra with Terraform and CI/CD. Designed ML feature pipelines, vector DBs, and monitoring systems. Improved DB performance, implemented security/compliance controls, and mentored engineers to accelerate delivery.

Kafka + Spark Structured Streaming with Avro and exactly-once processing improved reliability to 99%+ for 5M+ events/day. Optimized PostgreSQL and Redis reduced p95 latency by 30% and surfaced slow queries. Built OpenTelemetry tracing, Kafka monitoring, and DLQ alerts, cutting incident response time from 4h to 90m and improving system reliability and observability.

Built ML feature store supporting 10+ teams with Kafka, Spark, PostgreSQL CDC, and Redis caching, enabling 500M+ events/day with point-in-time correct features and versioning. Built Airflow/Spark framework adopted by 30+ engineers, cutting pipeline delivery time 60%. Improved Kubernetes CI/CD with Terraform and Jenkins, reducing on-call load from 15 to 4 hours/week.

Led Redshift to Snowflake migration using Terraform, improving query speed by 40% and reducing cost by 25% with clustering and materialized views. Built ingestion platform processing 50M+ records/day from multiple sources with idempotent pipelines powering real-time bidding. Implemented Datadog observability with SLOs, error budgets, and data quality monitoring.

Built claims ETL pipeline processing 170M+ insurance claims with 2x throughput using partition-aware incremental loading, parallel HDFS ingestion, and data validation. Designed HIPAA-compliant star schema models with AES-256 encryption, RBAC, and optimized Vertica projections, enabling sub-second queries and scalable analytics across large healthcare datasets.

Completed an M.S. in Computer Science at Stanford University from 2012 to 2014.

Completed a B.S. in Computer Science at Stanford University from 2008 to 2012.