Senior Research Scientist, Nemotron Post-Training

Details

Join NVIDIA and help build the Nemotron models that will define the foundation of open-source generative AI. We are looking for a research scientist / engineer who is passionate about open-source and excited to create our next-generation post-training pipelines. You will work at the intersection of research and engineering to invent, implement, and scale the core post-training technologies behind our Nemotron models.

What you’ll be doing

• You will be engaged as core contributors to Nemotron models post-training, working at the intersection of the areas:
  1. Synthetic data and algorithmic research for agentic RL
  2. Data and training Infrastructure implementation
  3. Collaborating in vendor data acquisition and experimentation
  4. Large-scale research & production model post-training
• Advance open-source foundation models by developing training data, benchmarks, LLMs and software (including NeMo-RL, Nemo-Gym and yet to be announced software)
• Solve large-scale, end-to-end foundation model post-training challenges, spanning the full model lifecycle from initial orchestration, data pre-processing, running of model training and tuning, to model deployment
• Publish and present your results at academic and industry conferences

What we need to see

• Master or PhD degrees in computer science, machine learning or other quantitative domains (or equivalent experience)
• 5+ year working or research experience in model mid-training / post-training, reinforcement learning and agentic systems
• Hands-on experience in data curation and model training for Agentic and Reasoning capabilities
• In-depth experience in using or developing inference and deployment environments such as vLLM, SGLang or TRT-LLM

Ways to stand out from the crowd

• Industrial experience in reinforcement learning for leading foundation models
• Experience in optimizing model quality from real-world traffic feedbacks