Yangjun Zhang

About Yangjun Zhang

SDE with 10+ years of experience in backend, large-scale distributed systems, and big data architecture. Proven expertise in optimizing ExaByte-scale distributed storage and building high-performance OLAP and AI/ML infrastructure. Expert in driving cloud migration, cost efficiency, and technical modernization.

Experience

1. Staff Software Engineer

   Uber

   Mar 2024 — Present

   Led HDFS optimization for Uber’s Exabytes-scale Lakehouse with 2 EB datasets across 20+ clusters. Drove exceptional cost efficiency, achieving $36 million in cost savings by implementing high-density storage servers. Served as the storage tech lead during critical batch-analytic incidents. Mitigated outages across HDFS control plane, data plane and storage clients, and upstream/downstream systems, including Presto, Spark, Airflow, Hive, Michelangelo(Uber’s ML Platform, including Ray), Flink, Kafka- Spearheaded building a distributed cache(Alluxio) to address cloud migration challenges. Architected and implemented a high-performance C++ cache client and server based on ArrowFlight, solving complex distributed systems challenges like data coherence, thread/memory leaks, zero copy optimization, and ETCD saturation- Building multi-modal lakehouse (MMDP) for Uber\'s business, including integration and optimization LanceDB, CPU based LLM inference and embeddings, and Deepseek 3FS- Key contributor to migrate Uber datalake storage to cloud: integrated and optimized GCS/OCI connector into Uber\'s storage ecosystem. Addressed the blocking issue - GCS egress saturation. identified the read-amplification in the GCS connector, and reduced the amplification by 50%.

Education

• Beijing University of Posts and Telecommunications

  Bachelor of Network Engineering, Computer Science

• Tsinghua University

  Master's Degree, Computer Science