AI Scientist - DeepTech (M/F/X)

Qui sommes-nous ?

Join a deeptech startup building Foundation Models for structured data (tabular & time-series) instead of yet another generic LLM.

As an AI Scientist, you will work on pre-training, fine-tuning and synthetic data modelling of large transformer-based models, at the core of a production-grade platform used by major enterprise and digital-native clients. You report directly to the scientific leadership and collaborate closely with a compact team of ML engineers and researchers.

Description du poste

💰 80–120 K€ gross / year

📍 Paris area (on-site several days / week)

⏳ Start: ASAP

🏠 Remote: 2 days / week after probation

🌍 Language: English fluent / French optional

🎓 You must have a PhD in Machine Learning to apply.

Join a deeptech startup building Foundation Models for structured data (tabular & time-series) instead of yet another generic LLM.

As an AI Scientist, you will work on pre-training, fine-tuning and synthetic data modelling of large transformer-based models, at the core of a production-grade platform used by major enterprise and digital-native clients. You report directly to the scientific leadership and collaborate closely with a compact team of ML engineers and researchers.

🎯 What you will do :

• Design and improve foundation models for structured data (tables, time-series) based on modern transformer architectures
• Pre-train, fine-tune and evaluate large-scale deep learning models in the cloud or on private clusters
• Build and maintain evaluation frameworks and metrics aligned with real customer use cases (classification, regression, forecasting, anomaly detection)
• Drive active learning strategies and training data optimisation: sample selection, dataset curation, synthetic data generation, robustness and transfer learning
• Stay on top of the latest ML research and propose model or training improvements that translate into measurable performance gains.
• Communicate research ideas to the scientific community and experimental design to the internal ML team (papers, talks, internal notes)
• Collaborate with ML engineers, data scientists and customers to deliver representation algorithms that unlock high-impact downstream applications.
• Run ad-hoc analyses to understand and debug model behaviour, failure modes and scaling properties

🧰 Tech stack & environment :

• ML / DL: Python, modern deep learning frameworks (PyTorch, DeepSpeed, SLURM-based training), strong focus on transformer architectures.
• Foundation Models: large-scale pre-training and fine-tuning on tabular and time-series data, with strong emphasis on reproducibility and scalability.
• Data: large structured datasets (Parquet, SQL, NoSQL), high-volume transactional and behavioural data.
• Infra: cloud and private clusters for distributed training, experiment tracking, CI for research code.
• Culture: research-driven, fast-moving, high ownership, small senior team, strong fundamentals over hype

Profil recherché

👤 Who you are :

• PhD in Computer Science, Machine Learning or related field with a focus on deep learning.
• 3+ years of hands-on experience training, fine-tuning and evaluating deep learning algorithms (especially Transformers) at scale (cloud or private clusters).
• Strong background in machine learning theory and practice, with a rigorous approach to reproducibility and experimental design.
• Proficient in Python and modern ML frameworks/tools (PyTorch, Sklearn, experiment management, Git).
• Comfortable working with large structured datasets and data pipelines (Parquet, SQL / NoSQL).
• Excellent communication skills in English, able to work cross-functionally with researchers, engineers and business stakeholders.
• Self-starter, autonomous, thriving in a fast-paced early-stage environment with high expectations and a strong excellence mindset.

Strong bonuses:

• You have a publication record in top-tier ML conferences or journals
• You have demonstrated experience in designing and running large-scale ML experiments (SLURM, Pytorch, Deepspeed).
• Demonstrated machine learning experience in one of the following: open-source activity, data science competitions.
• Track record of translating research into business impact
• Experience in developing and debugging in C/C++, Python

💰 Package & conditions :

• Salary: 80–120 K€ gross / year depending on seniority and track record.
• Location: Paris
• Remote: up to 2 days / week after probation
• Equity: stock options (BSPCE) to reflect your impact and align long-term incentives.
• Benefits: comprehensive health insurance, paid leave aligned with French standards, dynamic work environment with a focus on work-life balance.

🧪 Why it’s interesting :

• Work on frontier research in tabular foundation models, an AI segment still largely unsolved and estimated as a massive next frontier for enterprise AI.
• Join a well-funded deeptech backed by Tier-1 investors and founders from leading AI and SaaS companies.
• Ship research directly into production for high-profile customers in finance, commerce and other data-intensive industries

🧪 Hiring process ;

Step 1: Intro call (30 min) with scientific leadership to align on motivation, research interests and mission fit.

Step 2: ~2 hours of deep technical interviews (research track record, DL/Transformer expertise, experiments, code).

Step 3: Follow-up conversations with leadership / peers to validate collaboration style and long-term fit