NVIDIA video team is located in Shanghai, focuses on algorithm, architecture and HW design of video/CV solutions in all NVIDIA products, including datacenter, desktop, automotive, embedded system, etc. More than 80% of global internet traffic is video stream and it’s still growing, there’s a strong demand for reducing video bandwidth and increasing video quality. We are looking for architecture engineers to address this problem in a DL based approach.

What you’ll be doing:

• Develop DL based solutions for video pre/post-processing, compression, etc. Including algorithm study, implementation, performance tuning, and productization.
• Study new video compression/computer vision technology, specifications, papers, etc.
• Define the chip architecture of NVIDIA's next-generation video/CV solution, collaborate with ASIC team to deliver HW design.

What we need to see:

• Master Degree or above in Compute Science or Electronic Engineering
• Minimum of 2+ years’ experience in deep learning or video codec projects.
• Experience with deep learning training frameworks such as Tensorflow, Pytorch, Caffe.
• Good programming skill and C/C++ or python coding abilities.
• Fluent English and good communication skill.

Ways to stand out from the crowd:

• Project experiences on deep learning-based video pre/post processing, compression.
• Experience with video codec such as H264/AVC, HEVC, VP9, AV1, AV2 or VVC.
• Experience with coding agents, such as Cursor, Codex, and Claude code.
• Background with computer vision projects such as optical flow, stereo, super-resolution, frame rate up-conversion, etc.
• Experience with CUDA.