What You Will Do
- R&D world models for industrial safety/physical AI: multimodal perception, temporal prediction, hazard simulation, counterfactual reasoning, closed-loop simulation.
- Use diffusion/VAE/autoregressive models; fuse vision, point cloud, time-series signals for physics-constrained generation and evolution prediction.
- Deploy in anomaly warning, accident replay, digital twin, risk assessment; own effect, performance, cost.
- Track frontiers, lead design/review, grow the team.
What We Are Looking For
- Master's+ in CS, Automation, Robotics, Physics, Math, etc.; 5+ yrs algorithm R&D.
- Strong foundation in computer vision, multimodal understanding, time-series analysis, or world models
- Proficiency in Python and mainstream machine-learning frameworks, with the ability to move algorithms from validation to deployment.
- Solid DL; familiar with Transformer, generative models, multimodal fusion.