AI research intern - H1 2027

The role

You will conduct literature reviews of LLM routing approaches and design methods to select suitable language models. You will evaluate solutions on product data and collaborate with Research Engineers and Data Scientists.

What you will do

• Conduct a structured literature review of LLM routing approaches
• Design and implement a routing method that automatically selects the most suitable language model
• Evaluate your solution on real product data across quality, cost, and latency dimensions
• Collaborate with Research Engineers and Data Scientists to iterate on hypotheses
• Contribute to the R&D roadmap by documenting findings and presenting results

Required qualifications

• Final year of a Master's degree (Bac+5) in applied mathematics, from an engineering school or university
• Solid grounding in statistics, machine learning, and deep learning
• Familiarity with data preparation techniques and feature engineering
• Demonstrated interest in research
• Python and core data science libraries (pandas, scikit-learn, PyTorch, TensorFlow, Transformers)
• Machine learning & deep learning
• Data preparation & feature engineering
• Research analysis and synthesis
• Ability to communicate technical concepts clearly to non-expert audiences
• French: Fluent
• English: Professional proficiency - B2/C1

Benefits

• Opportunity to be offered a permanent contract at the end of the internship
• Workplace in the 16th arrondissement with breakfasts, sports and yoga classes, after-work drinks, game nights, and CSE benefits
• Tailored onboarding journey
• Gymlib pass
• Swile meal voucher card
• 50% transport coverage (Navigo pass, bike, etc.)

Before you apply

• AI-powered screening conversation with Emma must be completed within 3 days

About the company

Mirakl is a marketplace innovation company founded in 2012. It provides an operating system that includes enterprise marketplace solutions, AI-powered multichannel selling, retail media, and agentic commerce infrastructure.