Tian Lin
Staff Software Engineer at Google DeepMind
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolTian Lin is a Staff Software Engineer based in Mountain View with 7 years of experience driving on-device ML at Google and DeepMind, previously serving as tech lead for TensorFlow Lite modeling. He has productionized state-of-the-art mobile models across CV, NLP, speech and recommendation—shipping MobileBERT, EfficientNet-lite, EfficientDet-lite and early Transformer support in TF2—and led end-to-end pipelines from training to TFLite deployment. Tian’s hands-on open-source contributions include migrating and refactoring custom ops and metadata for tensorflow/tflite-support and improving mobile ML examples and build tooling in tensorflow/examples, reflecting deep expertise in Android, C++ custom ops, and model metadata for edge devices. A PhD-trained researcher with a background spanning ML theory, startup CTO experience, and venture diligence, he brings both product instincts and low-level systems skill to scale ML to constrained devices. An interesting thread through his career is blending research-grade algorithmic rigor with practical engineering: from bandit and online-learning toolkits to shipping real-world on-device recommendation systems.
7 years of coding experience
15 years of employment as a software developer
Double degree of B.A. Economics, Double degree of B.A. Economics at Peking UniversityPh.D. Computer Science, Ph.D. Computer Science at Tsinghua UniversityThe Chinese University of Hong Kong (CUHK)
English, Chinese
Github Skills (12)
machine-learning10
tensorflow-lite10
c-language10
metadata10
gradlew10
cprogramming-language10
android-development10
gradle10
bazel9
object-detection9
nlp8
python8
Programming languages (6)
Github contributions (5)
tensorflow/examples
Sep 2019 - Oct 2021
TensorFlow examples
Role in this project:
ML Engineer
ML EngineerContributions:2 releases, 68 reviews, 160 commits in 2 years
Contributions summary:Tian's commits primarily involve refactoring and improving the code structure of a smart reply app, using Bazel and Gradle. They are building custom operations for the app, including writing C++ code to build the custom ops, packing them into an Android Archive (aar), and using gradle to build the application. The code changes demonstrate an understanding of TensorFlow Lite, Android development, and mobile machine learning.
tensorflow
tensorflow/tflite-support
Sep 2020 - Mar 2022
TFLite Support is a toolkit that helps users to develop ML and deploy TFLite models onto mobile / ioT devices.
Role in this project:ML Engineer
ML EngineerContributions:11 commits, 3 comments in 1 year 6 months
Contributions summary:Tian migrated a custom operation, `RaggedTensorToTensor`, to the `tflite-support` repository. They then modified existing code to refactor the namespace structure and packed sentencepiece and `ragged_tensor_to_tensor` into respective sub-namespaces. Additionally, the user added support for score calibration within the object detector metadata and incorporated those changes into the metadata files. This involved modifications to the metadata writers, test files, and the core object detector code to include the calibration.
iot-devicesmachine-learningtflite-modelstensorflowtflite
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.