Menglong Zhu
Product Manager - Camera at DJI
Los Angeles, California, United States
Summary
🤩
Rockstar🎓
Top SchoolMenglong Zhu is a product-focused camera leader and former Google senior engineer with 12 years of experience building real-time mobile and edge computer vision systems. With a PhD from the University of Pennsylvania, he pioneered MobileNets and led realtime object detection work that shipped across Google products and influenced industry standards. At DJI he progressed from machine learning specialist to director of ML and now leads camera product direction, blending research-grade ML expertise with product execution. He contributes to major open-source efforts like TensorFlow Models—working on RNN-based object detection and improving core model utilities—underscoring his hands-on approach to deploying robust models. Based in Los Angeles, he combines deep academic foundations in vision and SLAM with pragmatic product instincts for camera and embedded AI features.
12 years of coding experience
12 years of employment as a software developer
Bachelor of Science Computer Science and Technology, Bachelor of Science Computer Science and Technology at Fudan UniversityPhD in Computer Science Computer Science, PhD in Computer Science Computer Science at University of Pennsylvania
Stackoverflow
126reputation
14kreached
2answers
0questions
Github Skills (8)
model-building10
object-detection10
lstm10
rnn-model10
tensorflow10
n10
python10
machine-learning9
Programming languages (1)
Github contributions (5)
tensorflow/models
Aug 2018 - May 2019
Models and examples built with TensorFlow
Role in this project:
ML Engineer
ML EngineerContributions:17 commits, 13 PRs, 7 pushes in 9 months
Contributions summary:Menglong primarily contributes to the `lstm_object_detection` directory, demonstrating involvement in recurrent neural network (RNN) based object detection models. Their work includes refactoring the `ssd_meta_arch` for improved constructor arguments, adding tests for `lstm_cells.py` and modifying it with features like outputting bottlenecked features, clipping states, and making other configuration adjustments. Additionally, the user made modifications to the input and configuration utilities, which is key to model construction and training.
deep-learningtensorflow
dreamdragon/PennAction
Apr 2014 - Dec 2021
Contributions:20 commits, 10 pushes in 7 years 9 months
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.