Internship - Data Scientist (F/H) - April 2025 - 6 months

Le poste

Join us at LVMH, the world leader in luxury goods, where over 75 exceptional Maisons produce high-quality products across five major luxury sectors: Wines & Spirits, Fashion & Leather Goods, Perfumes & Cosmetics, Watches & Jewelry, and Selective Retailing. With over 213,000 employees globally and revenues of 86.2 billion euros in 2023, LVMH offers a dynamic environment for growth and innovation.

As part of the LVMH House team, your mission will be to design and develop Learning, Innovation, and Transformation solutions to address Business, Leadership, and New Ways of Working challenges. Join our international team and contribute to strategic projects in the field of Learning & Development.

Internship Title: Reinforcement Learning from Human Feedback on Large Language Models and Autonomous Agents

The LVMH Data Science department, a global leader in luxury, develops Artificial Intelligence algorithms for the Group’s Maisons. As a pioneer in the luxury sector, LVMH is committed to delivering an exceptional client experience, where AI plays an increasingly significant role in personalization, efficiency, and innovation. The "Human-Centered AI" approach is critical to ensuring that developed technologies respect human values and enhance client interactions. Integrating human perspectives into these systems ensures that AI amplifies human skills rather than replacing them.

Within this framework, Reinforcement Learning from Human Feedback (RLHF) is a key focus area. Methods such as DPO (Direct Preference Optimization) have already been explored to align our models' behavior with human expectations and values.

LVMH now aims to push this further by improving Large Language Model (LLM) pipelines, particularly Chain-of-Thought (CoT) and Retrieval-Augmented Generation (RAG), as well as Autonomous Agents, using human feedback. A key challenge involves leveraging global feedback on overall generated results and reassigning relevant information to optimize the models. This requires applying advanced RL techniques to effectively connect feedback with precise adjustments.

References:

• Direct Preference Optimization: Your Language Model is Secretly a Reward Model
• Training language models to follow instructions with human feedback

• Exploring Large Language Model based Intelligent Agents: Definitions, Methods, and Prospects
• On the Impact of Fine-Tuning on Chain-of-Thought Reasoning
• Understanding the planning of LLM agents: A survey
• Agent Q: Advanced Reasoning and Learning for Autonomous AI Agents

• ARES: Alternating Reinforcement Learning and Supervised Fine-Tuning for Enhanced Multi-Modal Chain-of-Thought Reasoning Through Diverse AI Feedback

• Multi-turn Reinforcement Learning from Preference Human Feedback

Les missions

Join the LVMH Data Science team to deepen your expertise in Artificial Intelligence applied to the luxury sector. Your responsibilities will include:

• Researching the state-of-the-art scientific and technological advancements in RLHF, CoT, RAG, and LLM pipelines applied to autonomous agents.
• Developing and experimenting with innovative approaches to enhance these methods using human feedback.
• Designing advanced RL techniques to efficiently map global feedback to specific model adjustments.
• Testing and evaluating solutions in complex scenarios, such as multi-turn dialogues or retrieval-augmented generation tasks.
• Communicating your findings and contributing to scientific publications or internal presentations.

What You Will Learn:

• Master advanced Machine Learning methods and contribute your innovations.
• Understand the strategic and technological challenges associated with AI industrialization.
• Explore unique challenges in the luxury sector and their connection to AI.

Votre profil

• Master’s degree specializing in Machine Learning or Artificial Intelligence.
• Research or final-year internship.
• Strong skills in applied mathematics (optimization, linear algebra, probability).
• Proficiency in programming (Python, PyTorch/TensorFlow, etc.).
• Ability to communicate complex topics and excellent written and verbal communication skills in French and English.

Informations complémentaires

Restaurant tickets, 75% public transport reimbursement, discount rates at the gym, free electric bikes available, integration seminar.

The internship offer is available from April 2025, for a period of 6 months and is based in Paris 8ème.

LVMH recruits and recognizes all talents.