Lin Yuan
Engineering Manager at Databricks
Palo Alto, California, United States
Summary
🤩
Rockstar🎓
Top SchoolLin Yuan is an engineering manager and former tech lead with 11+ years building high-performance ML and large-scale systems, now leading enterprise AI platform work at Databricks from Palo Alto. He combines deep algorithm and systems expertise—from ASIC/FPGA design automation and VLSI tools to distributed deep-learning frameworks—with hands-on performance wins such as cutting Mask R-CNN training from 2 hours to 25.7 minutes and helping achieve record MLPerf times. A committed open-source contributor and committer to projects like MXNet and Horovod, he focuses on low-level tensor/memory optimizations and distributed training integrations. Lin’s background in both hardware-aware co-design and cloud-native training infrastructure gives him a rare ability to bridge chip-to-cloud performance tradeoffs and deployable ML in production.
11 years of coding experience
17 years of employment as a software developer
Ph.D. Computer Engineering, Ph.D. Computer Engineering at University of MarylandB.S. Electrical Electronics and Communications Engineering, B.S. Electrical Electronics and Communications Engineering at Xi'an Jiaotong UniversityHigh School Attached to Hunan Normal University
Chinese
Github Skills (22)
unit-testing10
c-language10
distributed-training10
python10
operation10
tensorrt10
memory-management10
mxnet10
machine-learning10
horovod10
deep-learning10
tensorflow10
performance-optimization10
tensor10
cprogramming-language10
Programming languages (6)
Github contributions (5)
apache/mxnet
Jun 2018 - May 2020
Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more
Role in this project:
ML Engineer
ML EngineerContributions:59 commits, 206 PRs, 163 pushes in 1 year 11 months
Contributions summary:Lin primarily contributed to the development and enhancement of the MXNet deep learning framework. Their work includes adding a temperature parameter to the Softmax operator, optimizing runtime performance, and incorporating the new functionality with unit tests. Additionally, the user fixed bugs in several operators, including "where" and "ctc_loss", and updated unit tests to test the modified and new functionality. This indicates a focus on improving the core functionality and reliability of the MXNet library.
pythonschedulerdataflowmutationdata-science
horovod/horovod
Jan 2019 - Feb 2020
Distributed training framework for TensorFlow, Keras, PyTorch, and Apache MXNet.
Role in this project:ML Engineer
ML EngineerContributions:8 commits, 15 PRs, 3 pushes in 1 year 1 month
Contributions summary:Lin contributed to the Horovod framework, primarily focusing on enhancing its MXNet integration. Their commits included removing unnecessary code in MXNet, simplifying the MNIST example with the Gluon API, and fixing a bug related to tensor object lifetime. They also added the `-hostfile` option to `horovodrun`, improving deployment flexibility, and made adjustments to the MXNet image net example.
mpikeras-tensorflowtrainingbaidutensorflow
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.