About this role
Design and execute training pipelines and optimizations for large-scale AIGC diffusion-based multimodal models across image and video, including data curation, training stability, and post-training alignment to improve output quality. Investigate and prototype emerging multimodal capabilities, contribute architecture and algorithm design, deliver production-grade implementations, and contribute to the academic community through publications.
Skills for this role
AIGCDiffusion modelsMultimodal large modelsImage generationVideo generationLow-light enhancementData curationTraining pipelinesTraining stabilityPerformance optimizationArchitecture designAlgorithm designPost-training methodsModel alignmentMultimodal understandingVisual content generationPyTorchPythonC++CUDAResearchAcademic publicationsProblem-solvingTeam collaborationCommunication skills