NVIDIA's Cosmos generative AI team is looking to solve the problem of driving high-impact execution in foundation model building, powering breakthroughs in video generation, intelligent agents, simulation, and synthetic data.
Requirements
- Hands-on experience running generative AI models—whether LLMs, diffusion models, or VLMs/VLAs—along with a working understanding of their inputs, outputs, failure modes, and performance bottlenecks.
- Comfortable working in Python, using tools like Hugging Face, PyTorch, Docker, or similar frameworks to evaluate models, run inference, and understand code-level artifacts (e.g., config files, training checkpoints).
- Familiarity with the full lifecycle of foundation model development: data curation, training, post-training (e.g., distillation, quantization), evaluation, and deployment.
- Experience leading open-source or externally-released model artifacts (e.g., GitHub, Hugging Face, licensing, model cards, compliance metadata) is highly valued.
- Strong understanding of AI licensing, model/data provenance, and traceability systems is a major plus.
Responsibilities
- Lead generative-AI model building, including model training, post-training, optimization, and release to internal and external users.
- Coordinate and drive open source releases on platforms like GitHub and Hugging Face, ensuring quality, traceability, and findability.
- Supervise the model lifecycle and its dependencies (e.g., datasets, checkpoints, licenses) and maintain compliance documentation.
- Enable internal NVIDIA teams and early adopters with early access to models and tools; bring together and synthesize feedback to advise roadmap and product iteration.
- Drive multi-functional execution across Research Scientists, Research Engineers, Product, Marketing and Legal, acting as the glue to ensure alignment and high velocity.
- Collaborate with Legal to ensure all datasets, code, and models are licensed for commercial use and open source distribution.
Other
- We are looking for an outstanding technical program manager with a consistent record driving sophisticated, cross-functional AI initiatives across research, engineering, and product teams.
- You will be the connective tissue across science, engineering, legal, and product functions, and translate technical realities into clear program achievements, risk assessments, and execution plans.
- Excellent communication and leadership skills are essential—you should be able to influence at all levels, from individual contributors to VPs.
- 12+ years of relevant experience, including at least 5 years leading technical programs in ML, AI, or applied research domains.
- Applications for this job will be accepted at least until December 28, 2025.
