Location: Boston,MA, USA
Job responsibilities:
1. Participate in data processing and model development to ensure that the training set can be optimized for different tasks by collecting data from real environment and simulator (multi-modal data, including images, video, audio, 3D point cloud, etc.);
2. Build a training platform based on K8S;
3. Responsible for the development and optimization of specialist policy models and future generalist policies to support multi-tasking robot applications.
Job Requirements:
1. Master degree or above, major in computer, artificial intelligence, software, robotics, etc., more than 2 years working experience in the field of robotics, including but not limited to imitation learning/reinforcement learning training, task planning, control and navigation algorithm research, etc
2. Experience in developing large models (LLM, VLM, etc.) and multi-modal models; familiar with deep learning framework; able to conduct large-scale data training and optimization, especially for robot application scenarios
3. Familiar with data processing and data annotation processes, experienced in processing data from different sources and formats, especially multi-modal data in robot operation tasks
4. Experience in publishing papers in top conferences and journals in related fields (such as NeurIPS, ICML, ICLR, CVPR, ECCV, ICCV, ACL, EMNLP, MLSys) or other similar level or above is preferred.