Responsibilities
1. Responsible for the preference alignment algorithm of CapCut's multimodal model, and use SFT/RLHF/POST Training and other technologies to model domain knowledge for video creation 2. Improve the instruction compliance ability of the enhanced model and security capabilities of the video creation agent large model, improve the ability of the Pre-trained Model in video creation, and build an industry-leading intelligent agent for video creation experts.
Qualifications
1. Master degree or above in mathematics, computer science, control science, software engineering, artificial intelligence and other related disciplines 2. Familiar with the basic knowledge of large models, and have the basic knowledge of large language model related training or reasoning 3. Familiar with LLM training or Fine-tuning methods, such as SFT/RLHF experience, or familiar with reinforcement learning (RL) concepts and in-depth understanding of PPO related algorithm knowledge 4. Solid Python or C++ programming skills, familiar with PyTorch, Tensorflow, Deepspeed, Megatron, vLLM and other large model training and reasoning technology stacks 5. Passionate about technology, pay attention to new research and papers on large models, and interested in the application of large models 6. Have the ability to solve problems independently, have good technical communication and collaboration skills, and be willing to promote the solution of problems in the project. Bonus points: 1. Participated in RLHF work related to large model projects 2. In-depth understanding of alignment algorithm engineering practice, able to optimize the training and pushing efficiency of RLHF corresponding framework 3. Understand the alignment of multimodal large models and understand the related content of step supervised learning.
