Missions

- Collaborate within the HN Lab of UAR 3598 in the field of artificial intelligence (AI) by developing knowledge (publications) and proofs of concept in generative AI, in the field of neural networks and different types of language models (large, medium: LLM and MLM) feeding the use cases of the LLMs4EU program and some of which will be used in ISIDORE 2030.
- Develop a methodology and knowledge production on the development and use of LLMs and MLMs for SHS research, taking into account existing knowledge (machine learning) for the LLMs4EU program.
- Contribute to the activities of the LLMs4EU program within the HN Lab, in particular through cooperation with knowledge management companies.

Activités

Based mainly on knowledge of ISIDORE/NAKALA data and ISTEX corpora, the post-doc will :
- Produce articles demonstrating the interest of AI and expressive representation modes (semantic graphs, time series, ontologies) in addressing SHS issues.
- Develop proofs of concept implementing this knowledge within the LLMs4EU project work program.
- Implement machine learning methods to propose new AI models.
- Carry out machine learning processes based on all the SHS research publications and data present in priority in ISIDORE and NAKALA, and which can be extended to other collected data that could advantageously help detect models that could feed a conversational agent within the framework of LLMs4EU's needs.
- To disseminate the results of this work in the form of articles, conferences and presentations.

Compétences

- Expertise/Experience in AI and Deep and Machine Learning (specifically methods around LLM and MLM) specialized in SHS
- Expertise in retrieval-augmented generation methodologies
- Fluent English or good written English
- Open-mindedness, teamwork, autonomy and ability to interact with other disciplines such as computer science and knowledge engineering.
- Interest in interdisciplinary research
- An interest in social robotics (human-robot interaction) would be a plus.

Contexte de travail

The person will be recruited as part of the European LLMs4EU project led by DataSpace ALT-EDIC (Alliance pour les Technologies des Langues).

The person will be assigned to the Research Infrastructure - IR ∗ Huma-Num, supported by the Unité d'appui et de recherche (UAR) 3598, whose main supervisory bodies are the CNRS and Campus Condorcet, and whose secondary supervisory body is the University of Aix Marseille. To find out more about the IR∗ Huma-Num's missions: https://www.huma-num.fr/quest-ce-que-l-ir-huma-num/

The person will be assigned to the HN Lab under the responsibility of the HN Lab manager. He/she will work transversally with the various Huma-Num divisions within a unit currently comprising 25 people based in Paris and Lyon.
The person recruited will be assigned to the unit's Paris branch (Campus Condorcet, Aubervilliers 93).
This position allows for one or two days of teleworking per week, in accordance with CNRS regulations.
Regular travel (at least once a month on average) to the unit's various sites and throughout France is to be expected.