Multimodal Brain AI Research Engineer

Inria Paris; Paris; On-site; EUR 42,000 - 54,000; Full time

Job summary

Inria Paris within ARAMIS at the Paris Brain Institute (ICM) invites applications for a research engineer. The role focuses on building reproducible pipelines that extract representations from foundation models across histology, spatial transcriptomics, and MRI, and organizing them for analysis.

You will work with a PhD student to extend two-modality alignment to three modalities, develop data loaders, evaluations, and reusable tools, and contribute to reports and publications.

Qualifications

• Strong Python programming skills with practical experience in PyTorch and deep learning.
• Knowledge of representation learning or self-supervised learning.
• Good software practices: Git, code reviews, testing, documentation, and using GPU clusters (SLURM).

Responsibilities

• Write data loaders and preprocessing for public datasets (e.g., HEST-1k, TCGA, Ivy GAP, Allen Brain Atlas).
• Extract embeddings from pretrained foundation models per modality and organize them in a common format.
• Implement evaluation protocols with train/test splits and robust metrics for few-pair settings.
• Extend alignment method to three modalities; analyze results against baselines.
• Package code with tests, docs, and notebooks for open-source release.
• Participate in team meetings and contribute to progress reports and papers.

Skills

Python programming

PyTorch

Deep learning

Representation learning

Self-supervised learning

Git

Code testing

Documentation

GPU clusters

Tools

PyTorch

Git

SLURM

Job description

Inria Paris within ARAMIS at the Paris Brain Institute (ICM) invites applications for a research engineer. The role focuses on building reproducible pipelines that extract representations from foundation models across histology, spatial transcriptomics, and MRI, and organizing them for analysis.

You will work with a PhD student to extend two-modality alignment to three modalities, develop data loaders, evaluations, and reusable tools, and contribute to reports and publications.