M/F Post-doctorant(e) PC4-WP2

CNRS; France; Sur place; EUR 42 000 - 54 000; Plein temps

Résumé du poste

CNRS, within the DATA TERRA program in France, invites applications for a Postdoctoral Researcher to develop a conversational agent for geology, focusing on descriptor selection using existing semantic resources.

The project spans ontologies, semantic alignment, HPC experiments, and collaboration with Earth science experts and Data Terra teams to deliver evaluable outputs and publications.

Qualifications

• Strong programming skills in Python.
• Experience with LLM-based applications or AI agents.
• Knowledge of Semantic Web technologies (RDF, OWL, SPARQL).
• Excellent written communication for diverse audiences.

Responsabilités

• Conduct a state-of-the-art review of ontologies and thesauri in solid Earth sciences.
• Contribute to semantic artifact requirements for PEPR use cases and REGEF stakeholders.
• Identify semantic resource alignment methods for the project.
• Identify the best query decomposition approach for this work.
• Develop the conversational agent based on three geology use cases.
• Design and implement experimental protocols on local HPC systems.
• Analyze results and contribute to method improvements.
• Identify and structure geological concepts from texts and databases.
• Automate updating and enrichment of geological terminologies.
• Collaborate with Earth science experts and Data Terra teams.
• Produce technical reports, documentation, and publications.

Connaissances

Python

LLM applications

Semantic Web

Written communication

Description du poste

Organisation/Company CNRS Department DATA TERRA Research Field Computer science Mathematics » Algorithms Researcher Profile First Stage Researcher (R1) Application Deadline 16 Sep 2026 - 23:59 (UTC) Country France Type of Contract Temporary Job Status Full-time Hours Per Week 35 Offer Starting Date 15 Oct 2026 Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Is the Job related to staff position within a Research Infrastructure? No

Offer Description

The objective of the postdoctoral project is to develop a conversational agent tailored to the field of geology, with the aim of assisting geologists in identifying the most appropriate descriptors for characterizing their samples based on existing semantic resources. In line with this specialization objective, it is essential to evaluate the outputs generated by the conversational agent.

Responsibilities

• Conduct a state-of-the‑art review of existing ontologies and thesauri across the various disciplines of the solid Earth sciences, including those related to the subsurface.
• Contribute to the analysis of semantic artifact requirements specific to the PEPR use cases (e.g., resource characterization, subsurface management) and to the needs for describing geological sample data identified by the REGEF Research Infrastructure and other stakeholders within the geology community, in order to define three use cases for experimentation.
• Identify semantic resource alignment methods that are best suited to the project context.
• Identify the most appropriate query decomposition approach for this work.
• Develop the conversational agent based on three use cases to be defined in collaboration with geologists.
• Design and implement experimental protocols on local high‑performance computing (HPC) systems.
• Analyze and evaluate the results, contributing to the continuous improvement of the approaches developed.
• Identify and structure geological concepts from textual corpora and existing databases.
• Automate the updating and enrichment of geological terminologies.
• Work closely with Earth science experts to ensure the scientific relevance of the developments, and collaborate with the technical teams of the Data Terra Research Infrastructure to ensure the integration and consistency of this work within the Data Terra framework.
• Participate in PEPR working groups and produce key project deliverables, including technical reports, documentation, and scientific publications.

The ANR PEPR "Subsurface" project – PC4: Digital Earth Platform aims to improve the processes for collecting, producing, and exploiting geological data and knowledge. Scientific advances in these areas continuously lead to the development of new tools and methodologies for data and knowledge acquisition, including 3D integration and multi‑scale, multi‑physics simulation.

Developing the most accurate possible understanding of the Earth's subsurface is central to major economic, environmental, and societal challenges. Achieving this objective requires an integrative approach, as geology relies on a wide range of complementary scientific disciplines. The ambition of the PEPR project is to establish a unique, multidisciplinary platform capable of providing a shared quantitative representation of the subsurface.

This common knowledge framework will be built around an integrated digital ecosystem comprising heterogeneous scientific datasets, models, analytical tools, and interoperable workflows. Such an ecosystem raises significant scientific and technical challenges, including the mutual understanding of data produced by different disciplines, the reuse of cross‑disciplinary datasets, and the seamless integration of advanced data analysis and data mining tools. The overarching objective of the platform is to address these scientific challenges and enable a new generation of integrated subsurface knowledge.

Within this framework, the FormaTerre initiative and BRGM collaborate closely as part of the PEPR "Subsurface: A Common Good" programme to improve the integration and interoperability of data originating from multiple subsurface‑related disciplines, including geology, hydrogeology, geophysics, and geochemistry. In this context, knowledge representation through ontologies plays a key role in enhancing data semantics, facilitating data interoperability, and supporting modelling, analysis, and decision‑making processes.

However, despite the availability of semantic resources, their effective reuse by geologists for describing geological datasets and physical samples remains a significant challenge for domain experts.

Required Skills

• Strong programming skills in Python.
• Previous experience in developing Large Language Model (LLM)-based applications or AI agents is highly desirable.
• Knowledge of, or experience with, Semantic Web technologies, including RDF, OWL, SPARQL, and logical reasoning, would be an asset.
• Excellent written communication skills, with the ability to communicate complex concepts clearly to diverse audiences.

Additional Assets

• Interest in Earth sciences and the ability to collaborate effectively with domain experts.
• Familiarity with existing ontologies and thesauri in the geosciences domain.
• Ability to work independently, with strong scientific rigor and a collaborative mindset suited to interdisciplinary research.
• Strong analytical skills, initiative, and a proactive approach to problem solving.