AI RESEARCH SCIENTIST

STATION F; Paris; Hybrid; EUR 90,000 - 130,000; Full time

Benefits offered by this job

Occasional remote

Job summary

Scienta Lab is seeking an AI Research Scientist to design, train and scale deep learning models that learn meaningful representations from large-scale biological data to accelerate drug development in immunology. You will push the EVA foundation model, tackling cross-species transfer learning, multimodal integration, and predictive modeling of drug response at the patient level.

Working at the intersection of deep learning research and computational biology, you will collaborate with

Qualifications

• PhD in Machine Learning, Computer Science, Applied Mathematics or related quantitative field.
• Hands-on experience training deep learning models on large datasets.
• Publication record in top-tier ML venues (NeurIPS, ICML, ICLR, AAAI, or equivalent).

Responsibilities

• Design, train and iterate on transformer-based foundation models for biological data at scale.
• Lead authorship on high-impact publications and represent the company at international conferences.
• Translate model predictions into actionable insights for drug development (target validation, patient stratification, biomarkers).

Education

PhD in Machine Learning or related quantitative field

Tools

PyTorch

Python

Large-scale data pipelines

Job description

About

Scienta is a deeptech company advancing translational research in immunology.

We develop EVA, the first AI foundation model purpose-built for translational prediction in immune-mediated diseases. EVA addresses one of the biggest challenges in pharma R&D: the preclinical–clinical gap, where fewer than 10% of immune drug programs successfully translate to humans. Our mission is to bring predictive clarity to this high-risk phase of development.

Built on more than 3 billion multimodal data points across 30 diseases and 70 tissues, EVA is designed to reason like an immunologist. Its biology-first architecture translates early experimental signals into predicted patient outcomes, with explainable outputs such as pathway-level activity and mechanism-linked biomarkers that directly inform R&D decisions.

Our approach is scientifically validated. In inflammatory bowel disease, EVA accurately replicated anti-TNF clinical trial outcomes using only preclinical and baseline transcriptomic data, outperforming existing academic and private models. Our work is backed by 18 peer-reviewed publications and 2 patents.

Located at BioLabs Hôtel-Dieu (Paris), Scienta is supported by leading investors, was selected among Station F’s Future 40, and named a laureate of the EIC Accelerator. Our team brings together AI, immunology, and pharma expertise, with a strong PhD backbone.

If you want to work on hard scientific problems with real impact, you’ll feel at home at Scienta!

Job Description

As a AI Research Scientist at Scienta Lab, you will design, train and scale deep learning models that learn meaningful representations from large-scale biological data to accelerate drug development in immunology and inflammation. You will push the boundaries of our EVA foundation model, a multimodal transformer trained on tens of millions of transcriptomic samples and histology images, tackling challenges such as cross-species transfer learning, multimodal integration, and predictive modeling of drug response at the patient level.

Working at the intersection of deep learning research and computational biology, you will tackle core modeling challenges: designing architectures that handle high-dimensional, sparse and noisy biological signals across transcriptomics, histology and clinical modalities; establishing scaling laws for biological foundation models; and building training pipelines that harmonize heterogeneous datasets across species, sequencing technologies and resolutions. You will collaborate with immunologists, computational biologists and ML engineers in a multidisciplinary environment, leading research projects that translate model predictions into actionable insights for drug development: target validation, patient stratification, biomarker discovery and clinical trial outcome prediction.

We value scientific rigor, innovation, and aim to publish breakthrough research in top-tier venues. This position offers exceptional opportunities for leadership, mentorship, and driving the scientific direction of our AI research program.

Main Missions

• Design, train and iterate on transformer-based foundation models for biological data, working with omics data, histopathology images and clinical variables at scale
• Develop and optimize pre-training strategies and fine-tuning approaches for few-shot transfer across species, tissues and disease contexts
• Lead authorship on high-impact publications in top-tier AI and computational biology venues, represent the company at international conferences and workshops
• Conduct scaling experiments, establish compute-performance trade-offs and drive architectural decisions informed by rigorous benchmarking on drug development tasks (target efficacy prediction, molecular perturbation modeling, patient stratification, treatment response prediction)

Preferred Experience

• PhD in Machine Learning, Computer Science, Applied Mathematics or related quantitative field, with hands-on experience training deep learning models on large datasets
• 3+ years of research experience in deep learning with a publication record in top-tier ML venues (NeurIPS, ICML, ICLR, AAAI, or equivalent)
• Familiarity with large deep learning model trainings, and associated experimentations frameworks
• Expert-level Python programming, strong experience with PyTorch, and practical familiarity with large-scale data pipelines

How To Stand Out

• You have high agency, look for solutions rather than problems and you are a team-player willing to go beyond your job description
• Prior hands-on work on high-dimensional biological data
• Experience with foundation models, self-supervised learning, contrastive learning, or multimodal architectures applied to scientific or biological data
• You have trained from scratch models with >300M parameters.
• You are experienced in working in dynamic, changing and ambiguous environments such as startups, with rapid iterations

Additional Information

Contract Type: Full-TimeLocation: Paris Occasional remote authorized