The objective of this 6 month internship  is to study creation of high
quality synthetic data to fine-tune agentic LLMs. The tasks will be
defined during the internship. You will be working on creating a
pipeline to create both synthetic instruction and reinforcement data.
You will then finetune LLMs on the synthetic dataset and perform
extensive evaluation. Theoretical and practical knowledge of training
language models with huggingface or similar libraries is essential to
carry out the internship.


*Purpose*

-   Internship supervised by the Dragon LLM (AI team)
-   Internship of 6 months in Paris (fully on-site)
-   Full time, 35 hours weekly
-   Starting date: early 2026


*Responsibilities*

1.  Study state-of-the art in Large Language Models (LLMs) synthetic
    data generation and agentic LLMs.
2.  Compare different methods through systematic experiments.
3.  Finetune models on the generated data with RL.
4.  Perform evaluations using lm-evaluation-harness library
5.  Publish results in a workshop or a conference.


*Qualifications*

-   Master's Degree student in Computer Science, Machine Learning or
    Natural Language Processing.
-   Experience with LLMs (class or personal projects) (this is a must)
-   Experience with Huggingface library (this is a must)
-   Proficient in English and/or French.

*Benefits*
-   Ticket restos
-   50% of your Navigo pass is reimbursed
-   Working in an innovative and supportive environment at the
    forefront of AI developments!

Supervisors:
Gaëtan Caillaut
Raheel Qader

To apply:

Please send your CV to :  hr@linguacustodia.com


About Dragon LLM

Dragon LLM is a French company specialised in frugal and sovereign AI
models & architectures. Winner of the Large AI Grand Challenge
organised by the European Commission, Dragon LLM designs efficient and
economical AI models, adapted to business practices and deployed on
local infrastructures. Based in Paris, the company aims to build a
useful, responsible and efficient European AI, at the service of
companies while promoting the technological sovereignty of the
continent.
www.dragonllm.ai

*Raheel Qader*