Tianshi Zhu, Developer in Shanghai, China
Tianshi is available for hire
Hire Tianshi

Tianshi Zhu

Verified Expert  in Engineering

Software Developer

Shanghai, China
Toptal Member Since
May 4, 2020

Tianshi loves optimizing distributed systems, big data pipelines, and real-time streaming applications. It thrills him when an out-of-memory issue is fixed or a Spark job becomes 50% faster, or a streaming application's throughput is increased by 10x. Tianshi looks forward to building the next reliable, scalable, and highly available distributed system.


Java, Flink, Google Cloud Platform (GCP), Spring 4, Risk, Machine Learning...
Redis, Apache Kafka, Hadoop, Scalding, Spark, Python, Java, Scala, ML Platform...
Python, Apache Pig, Apache Avro, Apache ZooKeeper, Apache Lucene, Apache Kafka...




Preferred Environment

G Suite, Zoom, Slack, MacOS

The most amazing...

...optimization I've done is a 100-line change that makes Redis transactions faster by 20x.

Work Experience

Principle Architect

2020 - PRESENT
  • Built the Risk infrastructure from the ground up and scaled the team from 0 to 36.
  • Built a real-time risk engine based on rules and models. This engine is reused across five products and processes millions of transactions per day.
  • Optimized fraud models and rules to lift the absolute success rate by over 5% and reduce the action rate from over 20% to less than 5%.
  • Designed and built a machine learning platform that supports self-serving model deployment, unified real-time and batch feature generation, dynamic model routing, and feature/model store.
Technologies: Java, Flink, Google Cloud Platform (GCP), Spring 4, Risk, Machine Learning, Cloud Deployment

Staff Software Engineer

2017 - 2020
  • Made the ML feature pipeline faster and more reliable which saved $3 million annually.
  • Optimized a data pipeline's performance by 20x so a product can be launched on time.
  • Rebuilt an online feature store based on a Redis cluster and Lua script that cuts latency from 100 milliseconds to 5 milliseconds (ms).
  • Led a team to design/implement a multivariate experimentation service that can handle 3,000 QPS (queries per second) per node and a client-side p50 of 5 ms.
Technologies: Redis, Apache Kafka, Hadoop, Scalding, Spark, Python, Java, Scala, ML Platform, Machine Learning, Cloud Deployment

Senior Software Engineer

2014 - 2017
  • Implemented a serialization system which reduced the p50 latency by 30% and CPU usage by 15%, and achieved $2.5M annual saving.
  • Mentored an intern to design and implement a prototype for facet search, and boosted exploratory search CTR by 15%.
  • Migrated 80% of Linkedin's search traffic from a legacy search system to the new APIs.
Technologies: Python, Apache Pig, Apache Avro, Apache ZooKeeper, Apache Lucene, Apache Kafka, Java

Member of Technical Staff

2013 - 2014
  • Developed an OpenStack Neutron plugin for Oracle Virtual Network Controller.
  • Built a UI to visualize and automate virtual network setup.
  • Developed an API back end to interact with OpenStack and virtual machines.
Technologies: OpenStack, Apache ZooKeeper, Java, C++

An ML Feature System That Supports Batch and Real-time Processing

An ML feature system that allows users to define the business logic in one place and compiled it to multiple back ends, including Spark, Scalding, and real-time streaming processing. The machine learning (ML) features are guaranteed to not have time-traveling issues.
2011 - 2013

Master's Degree in Telecommunications

University of Pennsylvania - Philadephia, PA, USA

2007 - 2011

Bachelor's Degree in Telecommunications

Fudan University - Shanghai, China


Scalding, Apache Lucene


Slack, Zoom, G Suite, Apache ZooKeeper, Apache Avro, Flink


Spark, Hadoop, Django, Spring 4


Scala, Java, Python, C++


Apache Kafka, Windows, Apache Pig, OpenStack, Amazon Web Services (AWS), Kubernetes, MacOS, Google Cloud Platform (GCP)


MapReduce, Functional Programming


Redis, Cloud Deployment, PostgreSQL


Big Data Architecture, Distributed Systems, ML Platform, Risk, Machine Learning

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.


Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring