PhD Research Intern - Predictive World Models

Xpeng Inc.; Santa Clara (CA); Hybrid; USD 55,000 - 83,000; Full time

Benefits offered by this job

Competitive compensation package

Snacks, lunches, dinners

Job summary

XPENG is seeking PhD research interns to develop world models that learn the dynamics of the physical world from large multimodal data. You will work with diffusion, flow‑matching, video tokenizers, and transformer backbones, advancing predictive models for driving and robotics.

The role requires a PhD focus in DL/vision, strong Python skills, and a minimum 12 weeks on-site in Santa Clara. Publication potential is supported through collaboration with our research team.

Qualifications

• PhD candidate with strong DL/vision/generative modeling focus.
• First-author publications at top venues or arXiv submissions.
• Hands-on experience with PyTorch or JAX and large-scale data.

Responsibilities

• Drive a focused research project on predictive world models from problem formulation to evaluation.
• Explore multi-view future prediction and action-conditioned/independent formulations.
• Extend models to multimodal latent spaces for 3D scene representations.

Skills

PhD candidate in engineering or CS

Top-tier publications

PyTorch or JAX experience

Strong Python programming

12-week on-site availability

Education

PhD in Engineering/CS

Tools

PyTorch

JAX

Diffusion models

Flow matching

Gaussian Splatting

Transformer backbones

Job description

XPENG is seeking PhD research interns to develop world models that learn the dynamics of the physical world from large multimodal data. You will work with diffusion, flow‑matching, video tokenizers, and transformer backbones, advancing predictive models for driving and robotics.

The role requires a PhD focus in DL/vision, strong Python skills, and a minimum 12 weeks on-site in Santa Clara. Publication potential is supported through collaboration with our research team.