Foundation Models Engineer: Structured Data & Modern ML

NVIDIA Corporation; France; On-site; EUR 110,000 - 170,000; Full time

Job summary

NVIDIA Corporation in France is seeking a seasoned ML researcher to contribute to the next generation of structured data foundation models. You will collaborate with researchers and engineers to advance models over relational schemas, graphs, tabular data and time series, and design benchmarks to evaluate generalization and drive improvements.

This role covers the full ML lifecycle from architecture design and training to post-training optimization and inference acceleration, with a focus on

Qualifications

• MS or PhD in ML, CS or equivalent program.
• Proficiency in Python and PyTorch.
• At least 5 years of research experience in designing ML algorithm solutions.
• Practical experience in using Predictive Models in Real World Applications.

Responsibilities

• Contribute to our next generation of structured data foundation models.
• Collaborate with researchers/engineers to enhance foundation models over relational schemas, heterogeneous graphs, tabular data, and time series.
• Design and run benchmarks that evaluate generalization across structured data domains, driving model improvements.
• Leverage ML/AI knowledge to tackle real challenges with scalable and adaptable solutions.
• Work across the full lifecycle of modern ML systems: architecture design/training to post-training optimization and inference acceleration.

Skills

Python

PyTorch

ML research

Time series forecasting

Education

MS or PhD in Machine Learning or Computer Science

Tools

Chronos

TimesFM

TabICL

TabPFN

Job description

NVIDIA Corporation in France is seeking a seasoned ML researcher to contribute to the next generation of structured data foundation models. You will collaborate with researchers and engineers to advance models over relational schemas, graphs, tabular data and time series, and design benchmarks to evaluate generalization and drive improvements.

This role covers the full ML lifecycle from architecture design and training to post-training optimization and inference acceleration, with a focus on