Data Scientist

You'll be part of our Catalog Data Science team led by Arthur Delaitre and Adrien Morvan. As part of our broader Data team (60+ people), you'll be prototyping, iterating, and shipping algorithms to production that directly impact marketplace catalog challenges—from NLP to large-scale Generative AI with custom LLMs. Our opportunity can be located in Paris or Bordeaux and requires 4 days onsite per week, 1 day remote. Your Impact Build and deploy ML algorithms to production that power 500+ e-commerce and marketplace sites across 40 countries, directly impacting revenue growth, operational efficiency, and transaction safety Tackle real-world catalog challenges including automatic content rewriting, product attribute extraction from images and text, variant detection, product categorization, seller onboarding automation, and trending product prediction Work with cutting-edge AI techniques including multimodal models and LLM fine-tuning—Mirakl is one of the few French players with fine-tuned LLMs in large-scale production Own your projects end-to-end: from data analysis and prototyping to production deployment with Data Engineers and dev teams, plus building dashboards to monitor algorithm performance Collaborate across teams to refine use cases, user experience, and integration paths while presenting results at weekly data science meetings What You'll Bring to the Role Experience: 6+ years of experience as a Data Scientist with strong hands-on NLP and applied ML in industry Proven track record of deploying Machine Learning algorithms to production Experience with Spark development for large-scale data processing Skills: Expertise in NLP and Computer Vision algorithms and state-of-the-art architectures (e.g., Transformers) Proficiency in Python and TensorFlow and/or PyTorch Knowledge of the latest LLMs and fine-tuning techniques Data-driven, pragmatic, and business-oriented approach Strong ownership and autonomy with excellent team collaboration Tech Stack Core ML/AI: Python, TensorFlow, PyTorch, Hugging Face LLM-specific: Autotrain, Unsloth, Galileo, LangChain, Anyscale Data infrastructure: Databricks, Spark, AWS (Amazon Redshift, S3), SQL, Airflow, Delta Lake