LLM Reinforcement Learning Framework Engineer at NVIDIA

Shanghai, Shanghai, China -

Full Time

Start Date

Immediate

Expiry Date

23 Jul, 26

Salary

0.0

Posted On

24 Apr, 26

Experience

2 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

Python, PyTorch, Reinforcement Learning, Large Language Models, Distributed Training, Multi-GPU, NeMo RL, Megatron-LM, DeepSpeed, vLLM, TensorRT-LLM, Asyncio, Torch.distributed, Ray, GPU Architecture, Performance Optimization

Industry

Computer Hardware Manufacturing

Description

NVIDIA has been transforming computer graphics, PC gaming, and accelerated computing for more than 25 years. It’s a unique legacy of innovation that’s fueled by great technology—and amazing people. Today, we’re tapping into the unlimited potential of AI to define the next era of computing. An era in which our GPU acts as the brains of computers, robots, and self-driving cars that can understand the world. Doing what’s never been done before takes vision, innovation, and the world’s best talent. As an NVIDIAN, you’ll be immersed in a diverse, supportive environment where everyone is inspired to do their best work. Come join the team and see how you can make a lasting impact on the world. As a key player in the AI revolution, NVIDIA is pushing the boundaries of what's possible in accelerated computing. We are seeking an exceptionally dedicated LLM Reinforcement Learning Framework Engineer to join our ambitious team. This role is vital in advancing our large language model (LLM) capabilities, particularly in improving reasoning abilities for math, coding, and agentic AI. Join us and contribute to groundbreaking innovations that will craft the future of computing! What you’ll be doing: Developing and deploying reinforcement learning algorithms for LLM post‑training to improve reasoning and alignment. Integrating RL components into NVIDIA’s LLM training and serving stack with a cross‑functional team of engineers and researchers. Crafting and running experiments, evaluations, and debugging workflows to ensure robustness, scalability, and reliability in production. What we need to see: Strong Python programming skills with production‑quality PyTorch experience in multi‑GPU and distributed training environments. Hands‑on experience with modern LLM frameworks such as NeMo RL, Megatron‑LM, DeepSpeed, vLLM, TensorRT‑LLM, or similar is a big plus. Practical experience with reinforcement learning applied to LLMs or large‑scale sequence models. Familiarity with async and distributed orchestration (e.g., asyncio, torch.distributed, Ray, or equivalent). 3+ years of relevant industry or research experience, and BS/MS (or equivalent) in CS, CE, EE, or a related field. Solid foundations in probability, optimization, statistics, and deep learning. Understanding of GPU architecture and performance optimization is a strong plus. Strong problem‑solving, debugging, and collaboration skills, with a passion for innovation and delivering industry‑leading AI solutions. Join us at NVIDIA and help build the next generation of reasoning‑capable LLMs that make a lasting impact on the world! Widely considered to be one of the technology world’s most desirable employers, NVIDIA offers highly competitive salaries and a comprehensive benefits package. As you plan your future, see what we can offer to you and your family www.nvidiabenefits.com/ NVIDIA is committed to fostering a diverse work environment and proud to be an equal opportunity employer. As we highly value diversity in our current and future employees, we do not discriminate (including in our hiring and promotion practices) on the basis of race, religion, color, national origin, gender, gender expression, sexual orientation, age, marital status, veteran status, disability status or any other characteristic protected by law. NVIDIA pioneered accelerated computing. Today, our AI infrastructure powers global intelligence, transforming every industry. Learn more about NVIDIA.

Responsibilities

Develop and deploy reinforcement learning algorithms for LLM post-training to enhance reasoning and alignment capabilities. Integrate RL components into the existing training and serving stack while conducting experiments and debugging workflows for production reliability.