1. Research and develop sate of the art deep learning compression including model distillation, pruning, quantization, and others for CNN, RNN and LSTM graphs.
2. Use deep learning frameworks such as TensorFlow, Caffe2 or PyTorch to develop advance training algorithm to auto prune and reach low-bit quantization.
3. Apply model computation reduction techniques to computer vision applications and others.
4. Close interaction with System, Hardware and Software architects to best leverage the features of the CNN engine and run time.
5. Ability to travel internationally up to 5% of the time to work with partners.
1. Master degree in Engineering, Computer Science, Imaging Science, related field or equivalent practical experience.
2. Used to work in computer vision and machine learning with a focus on deep learning.
3. Solid understanding of CNN, RNN, LSTM and their training using hyper-parameters tuning.
4. Familiar to code with Python or C++.
5. Experience with SW development processes (project planning, version control, bug tracking)
6. Excellent teamwork and communication skills.
7. Able to speak and write English fluently.
8. Has relative experience of image recognition.