Research Scientist Foundation Models and Multimodal AI

Il y a 3 jours

Paris, Nouvelle-Aquitaine, France present of AI Temps plein 90 000 € - 110 000 €/an

Research Scientist Foundation Models and Multimodal AI

The role centers on taking existing open foundation models and adapting them to biological and clinical data rather than building new base models. A strong candidate has a few years of hands on deep learning work, comfort fine tuning models, and experience joining different data types such as tables, gene expression, images and text into shared representations. Solid Python and PyTorch skills plus careful evaluation habits matter as much as research curiosity.

How it fits the AI era

The work sits squarely in the AI era, since the daily task is shaping and aligning pretrained models for a scientific product. Much of the routine tuning and benchmarking can be sped up with tooling, so judgment about which adaptations are worth doing and how to measure them stays with the person. Understanding the biology and asking the right research questions remain human contributions.

How it builds AI experience

A person here would gain practical skill in fine tuning and aligning open models across several data types, and in building evaluation pipelines for scientific use. Working next to a company that pairs an AI scientist with pharmaceutical research teams offers exposure to how models get used in real discovery settings. Building this background could come from hands on projects with multimodal or omics data and from publishing or sharing reproducible work.

Worth showing

  • Multimodal embedding alignment
  • PyTorch and Hugging Face
  • Benchmark and evaluation design
  • Omics or clinical data work

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Owkin

Per the company, Owkin builds K Pro, an AI scientist that works alongside pharmaceutical research teams to speed discovery and strategic decisions.

Tools mentioned: machine learning, deep learning, Python, AWS, GCP, Kubernetes