Dong Nie

Work on post-training, test-time scaling for large models, llm for personalization. Also have experience in embedding for personalization and watermarking for fingerprinting.

Lead a team (25 great engs) working on perception for intelligent driving, including data, front/rear/side view perception, fish-eye perception, novel AV algorithms, tracking/fusion, deployment, etc. Also work on genAI for multi-modality based products, e.g., conversation based navigation and image editing, including collect data, clean data, fine tune the model, evaluation and help deployment to vehicle.

Work on perception for mapping and intelligent driving.

I work on recognition of large scales of real-world data, in particular, perception for mapping.

My research interests are machine learning based medical image analysis and text mining, especially focus on adversarial confidence learning for medical image semantic and instance segmentation and synthesis.

CNN System using OpenGL ES on Android Platforms. Specifically, Implement the basic CNN-related operations by shaders for parallel computation. Write C++ codes to call the shaders and encapsulated as a whole system. Develop the interface with Jave which call the C++ components with JNI.

Machine Learning in Medical Image Analysis, such as adversarial confidence learning for medical image segmentation and synthesis.

Machine learning for web mining: mental health mining based on social media behavior. Movie recommendation using unrated data and personality information.

Learning basic courses for computer science, including C/C++/JAVA/web programming/Operating System/Computer Network/Computer Architecture/Compiler/Data Structure/Algorithms.