Tasks:
- * Analyze trajectory smoothness and task success
- * Assist with model training and fine-tuning
- * Build robot action post-processing modules
- * Deploy and validate VLA models on robots
- * Design action quality metrics
- * Develop action smoothing, retiming, and speed adaptation
- * Document experiments and technical findings
- * Implement latency compensation and execution stability improvements
- * Iterate with modeling, data, control, and engineering teams
- * Optimize VLA action chunks and trajectories
- * Process robot data and analyze action trajectories
- * Research real-time inference and action acceleration
- * Run offline evaluations and real-robot experiments
Perks/Benefits:
Skills/Tech stack required:
[Action Chunking] [CUDA] [Deep learning] [Diffusion Policy] [Flow matching] [Imitation Learning] [Inference Optimization] [Linux] [Model Inference] [Model Inference Optimization] [Motion smoothing] [MPC] [Multimodal Models] [Python] [PyTorch] [Real Time] [Real-time Inference] [Robot arm] [Robot Arm Control] [Robot Learning] [ROS] [ROS 2] [TensorRT] [Trajectory Interpolation] [Trajectory Optimization] [Transformer-based] [Transformer-based policies] [Triton] [Vision-language] [Vision-language-action] [Vision Language Action VLA] [VLLM]
Educational requirements:
[Bachelor's Degree] [Master's Degree]
Role(s):
[Algorithm Intern] [Intern] [Robotics Algorithm Intern] [VLA Algorithm Intern]