NVIDIA

We are looking for interns in three different areas: CVCG & Simulation, LLM Applications, Accelerated Computing. This role involves collaborating with the research and product teams to address algorithm optimization challenges and drive the deployment of NVIDIA solutions for customers. As an intern, you will have the opportunity to work on cutting-edge projects that advance the state of the art in AI and high-performance computing. You will also gain hands-on experience by contributing to real-world solutions, helping to bridge the gap between innovative research and impactful customer applications.

What you’ll be doing

• Familiarity with NVIDIA’s key products and technologies in autonomous driving and robotics, especially in 3D reconstruction/generation, world modeling, VLM, and VLA.
• Conduct feasibility studies, engineering implementation, algorithm optimization, and acceleration for specific technologies. Collaborate closely with NVIDIA’s research and engineering teams at headquarters to validate products/algorithms, optimize performance, and assist in resolving technical issues during customer deployment.
• LLM applications: Continuously track the latest advancements in LLM, Automatic Speech Recognition (ASR), Text-to-Speech (TTS), and multimodal fusion technologies within the industry and at NVIDIA. Conduct cutting-edge algorithm research and optimization, including but not limited to: Model architecture improvements Training efficiency enhancement Reinforcement learning techniques Multimodal alignment strategies Conduct algorithm validation, model training, and inference optimization for LLM, ASR/TTS, and multimodal fusion models. Explore novel application paradigms (e.g., Agentic AI) and develop end-to-end engineering solutions.
• Accelerated Computing: Developing and optimizing computational performance for distributed training. Continuously exploring methods and technologies such as low-precision training, sparse computation, and parallelism strategy tuning to enhance performance. Working on inference frameworks by optimizing popular frameworks (e.g., SGLang, TRTLLM), adding new features, and resolving performance bottlenecks. Optimizing deep learning frameworks (e.g., GNN training frameworks) through system design, parallel algorithms, and network communication improvements, tailoring code for specific architectures and algorithm characteristics.

What we need to see:

• CVCG &Simulation: Experience in autonomous driving or robotics is preferred; Research or development experience in 3D reconstruction, world modeling, VLM, or VLA; Strong programming skills, fast learning and adaptability, with the ability to independently analyze, define, and solve problems.
• LLM applications: Research or development experience in LLM, multimodal AI, or speech algorithms, with familiarity with mainstream architectures (e.g., Transformer, Mixture-of-Experts). Hands-on experience in LLM training or ASR/TTS technologies, proficiency in deep learning frameworks (e.g., PyTorch). Strong programming skills; familiarity with distributed training/inference frameworks (e.g., DeepSpeed, Megatron, TRT-LLM) is preferred. Master’s degree or higher in Computer Science, Artificial Intelligence, NLP, or related fields, with solid theoretical foundations in machine learning.
• Accelerated Computing: Familiarity with one or more of the following: accelerated computing, parallel computing, distributed training, or inference, with a strong desire to deepen expertise in these areas.
• Master’s degree or higher in Electronics, Automation, Computer Science, Computational Mathematics, or related fields.
• Strong programming skills with solid understanding of data structures and fundamental computer system concepts.

Ways to stand out from the crowd:

• Ability to independently solve complex technical problems and passion for continuous learning.
• Contributions to relevant Open Source Software projects.
• Published research work on relevant domains