AI / Algorithm Intern – CV · Multimodal · LLM
You will join 51WORLD’s core AI organization in Beijing or Shanghai and ship
production-grade research and engineering on top of one of the largest digital-
twin and synthetic-data platforms in the world. Based on your background, you
will be matched to one of the following tracks:
• Computer Vision & 3D Reconstruction — image/video understanding, NeRF,
3D Gaussian Splatting, large-scale scene reconstruction
• Multimodal AI — vision-language models, scene understanding, agent perception
• Large Language Models — pre-training, post-training, RLHF / RLAIF, evaluation,
agentic frameworks for simulation and decision-making
• Synthetic Data & Sim2Real — data generation pipelines, domain randomization,
closed-loop training for autonomous driving
• Reinforcement Learning & Decision Intelligence — planning, control and
multi-agent simulation in digital-twin environments
What You Will Do
• Own a well-scoped research or engineering problem end-to-end, from literature
review and prototyping to ablation studies and deployment.
• Train, fine-tune and evaluate models on multi-GPU clusters (NVIDIA A100 / H100).
• Build evaluation pipelines and contribute to internal benchmarks.
• Collaborate with senior researchers, product managers and platform engineers
to ship features into 51WORLD’s commercial digital-twin and simulation products.
• Co-author internal tech reports and, where the work qualifies, top-tier
conference submissions (CVPR / NeurIPS / ICLR / ICCV / ICML / SIGGRAPH).
Minimum Qualifications
• Currently enrolled in a Bachelor’s, Master’s or PhD program with an expected
graduation date in 2027.
• Major in Computer Science, Artificial Intelligence, Electrical Engineering,
Applied Mathematics, Statistics, Robotics or a closely related quantitative
field.
• Strong programming in Python and hands-on experience with PyTorch.
• Solid foundations in deep learning, linear algebra, probability and
optimization.
• Available and willing to work ON-SITE in Beijing or Shanghai for at least
4 months. This role is not open to remote / overseas-based work — candidates
studying abroad must be able to relocate to mainland China for the internship.
Preferred Qualifications
• First-author publications or accepted papers at top venues (CVPR, ICCV, ECCV,
NeurIPS, ICLR, ICML, ACL, EMNLP, SIGGRAPH, RSS, CoRL, etc.).
• Hands-on experience with one or more of: LLM training/fine-tuning, diffusion
models, NeRF / 3D Gaussian Splatting, multimodal pre-training, RLHF, CUDA,
large-scale distributed training (DeepSpeed / Megatron / FSDP).
• Open-source contributions to mainstream AI / 3D / simulation projects.
• Prior internship at a leading AI lab, autonomous-driving company or research
institute.
• Working proficiency in English; Mandarin is a strong plus given the on-site
collaboration setting.
