Drimva

NVIDIA

Principal Research Scientist, Synthetic Data Generation

$272K–$431K

US, CA, Santa Clara +4 · Principal · Full-time · Posted Sep 4, 2026

Location
US, CA, Santa Clara
US, CO, Remote
US, CA, Remote
US, NY, New York
US, MA, Remote
Experience
15+ years in engineering and research in synthetic data generation, generative modeling, multimodal machine learning, or related areas.
Education
PhD in Computer Science, Machine Learning, Statistics, or a related field, or equivalent experience.

About the job

Sets technical direction and builds open-source tools and scalable pipelines for generating and evaluating synthetic data used to train frontier language and multimodal models.

What you'll do

  • Build synthetic data pipelines for model pre-training, post-training, reasoning, coding, structured outputs, and multimodal tasks.
  • Develop agentic, tool-use, reinforcement-learning, privacy-preserving, and safe data-generation methods.
  • Maintain open-source libraries and SDKs; publish research and mentor scientists and engineers.

What they ask for

  • PhD in Computer Science, Machine Learning, Statistics, or a related field, or equivalent experience.
  • Deep understanding of LLM training stages and inference frameworks such as vLLM or TGI.
  • Track record developing or maintaining software libraries used by a broad developer community.
  • Experience building scalable data pipelines for large-scale model training; strong publication record at premier venues.

Nice to have

  • Significant open-source contributions with community adoption.
  • Experience in multimodal generation, agentic or reinforcement-learning data, differential privacy, or regulated-industry synthetic data.

Pay

272,000 USD - 431,250 USD (base salary)

Tools

NVIDIA NeMo, Nemotron, vLLM, TGI, Git, CI-CD

AI focus

Research and engineering for synthetic data generation for LLM and multimodal model training, including agentic and tool-use training, reinforcement learning, automated quality evaluation, privacy-preserving synthesis, and safety.

Model-written summary · the original posting is authoritative

First seen Oct 15, 2026 · Last verified Nov 5, 2026 (UTC)