Overview In this role you advance agentic multimodal AI within NVIDIA's Nemotron LLM team. You apply foundational and applied research to build and refine language-capable neural models that reason, plan, and act across audio-visual modalities. You will shape core algorithmic improvements for multimodal foundation models and work on high-impact large language models that touch millions of users. This is an opportunity to contribute to state-of-the-art AI computing at scale, in a collaborative, innovation-driven environment.
Compensation / Benefits- equity
- comprehensive benefits package
- remote work option
- competitive salary
- career growth opportunities
- industry-leading salary range 184,000 USD - 287,500 USD
Responsibilities- Develop, train, fine-tune, and deploy advanced neural networks for language processing in agentic systems with audio-visual reasoning, tool usage, and document understanding
- Advance post-training and alignment methods including instruction tuning, preference optimization, and RLHF/RLVR for multimodal agents
- Research agentic reasoning and grounded perception with focus on planning, tool execution, and long-horizon tasks across digital and physical environments
- Lead collection, development, and benchmarking of multimodal datasets with quality evaluation of accuracy, safety, and task completion
Key requirements- Master's degree (or equivalent experience) or PhD in Computer Science, AI, or Applied Math with 8+ years of relevant experience
- Excellent Python programming with scalable model development and PyTorch experience
- Strong knowledge of ML/DL techniques and foundation model architectures (Transformers, Mixture-of-Experts)
- Foundational understanding of reinforcement learning algorithms (MDPs, policies, reward design)
- Hands-on experience with post-training multimodal models for audio-visual reasoning and human-AI interaction
- Proven ability to manage model development lifecycle (dataset versioning, experiment tracking, evaluation pipelines)
- strong collaboration and cross-functional communication
- problem solving and proactive experimentation
- ability to work independently and manage complex projects
- PyTorch
- Transformers
- mixture-of-experts models
