Hedra

As a Research Engineer on our Physical AI team, you will lead pre-training and post-training on action-conditioned world models, working hand-in-hand with industrial partners to close the loop between generative AI and physical systems. Responsibilities:Design, implement, and run pre-training and post-training pipelines for action-conditioned world models and vision-language-action (VLA) modelsDevelop and refine training methodologies, including fine-tuning, reinforcement learning, and large-scale multimodal learningDesign and generate training and evaluation datasets from simulation, including environment setup, domain randomization, and sim-to-real transfer strategiesBuild distributed training infrastructure using PyTorch, FSDP, and DeepSpeedWork with multimodal data pipelines involving video, sensory inputs, and action sequencesEvaluate model performance using both benchmark datasets

The RoleAs Senior Video Producer at Hedra, you'll own our video engine end to end. Source, contract, and manage the freelance bench of video editors, motion graphics designers, and on-camera talent. Today our API serves both first- and third-party models to production customers, and our agent works within a single workflow to create content from scratch, manipulate existing assets, critique its own outputs, and repurpose them on demand. He is co-first author on BEHAVIOR, a landmark embodied-AI benchmark, and has published at top conferences including CoRL, ICRA, and ICML (with a CoRL 2022 Best Paper nomination), with prior experience on NVIDIA robotics and simulation.