Employment
PhD
Vacancies at FEMTO-ST

You are here

DISC :"AI agents for healthcare applications"

Context : After the numerical revolution that was triggered by generative AI which was based on Large Language Models (LLMs) [1] and initiated by OpenAI and its ChatGPT chatbot, AI agents [2] are expected to be the next big technology in AI and it was the top trending technology in 2024 according to Gartner. AI agents are also based on LLMs but have more capabilities such as decision-making, problem-solving, interacting with external environments and executing actions. Contrary to LLMs, AI agents’ response is not limited by the data used to train them. To answer complex requests to which they do not have the full knowledge base, AI agents can perform task decomposition and create a workflow of specific subtasks. These subtasks might be solved using external tools that the AI agent can call. These tools can include external datasets, web searches, APIs, specialized AI models and
even other AI agents. After each subtask, the agent can update its knowledgebase, reassess its plan of action and modify it if necessary. Moreover, AI agents can improve their reasoning and accuracy
by learning from the feedback returned by other AI agents and humans. This will allow them to better
align their results with the users’ intended goals and avoid repeating the same mistakes by storing
solutions to previous obstacles in a knowledge base..[...]

As a case study, the proposed approaches in this PhD thesis will be applied to the pre-authorization in healthcare insurance. It represents a significant administrative burden for healthcare providers, resulting in delays in patientcare, increased operational costs, and clinician burnout. Healthcare insurance pre-authorization processes suffer from several critical issues:
• Time-consuming manual review of patient records and insurance policies, creating significant bottlenecks.
• Inconsistent interpretation of coverage criteria across reviewers, leading to systemic issues requiring internal reforms and enhanced training programs.
• High error rates in clinical documentation and coding, which automated systems like InsightRT have shown promise in reducing by integrating with electronic medical records [3].
• Delays in patient care due to administrative processing time, with orthopedic practices seeing mean approval times extending over 30 days for procedures like total hip and knee arthroplasties [4].
These challenges result in approximately$31 billion in annual administrative waste in the US healthcare system alone (American Medical Association, 2023), with 92% of physicians reporting that preauthorization processes negatively impact patient clinical outcomes.

Application deadline : 12/05/2025
Start date : 01/10/2025
Gross Salary : 2200€

 

https://www.femto-st.fr/en/job-training-thesis/phd#job-5768

DISC :"Optimization of drone flight plans for the collection of agriculture data"

Description : Wireless Sensor Networks (WSNs) and robotics are revolutionizing the agricultural industry. On one hand, WSNs provide valuable environmental information (such as temperature, soil conditions, crop imagery, etc.). On the other hand, robots act based on this information, for instance, for mechanical weeding, seeding, or harvesting tasks. However, data collection techniques for agriculture still need improvement. This project proposes the use of a Wireless Underground Sensor Network (WUSN) to avoid interference with the operation of agricultural machinery. [...]

This PhD is part of the ANR-funded OCOD project led by INRAE Clermont-Ferrand. The project brings together the expertise of four laboratories — TSCF, FEMTO-ST, LPCA, and Inria-FUN — to tackle challenges related to data acquisition in natural agricultural environments that are difficult to access, using aerial data collection platforms.

Applicant Profile :  The candidate must have a strong background in combinatorial optimization, particularly in integer linear programming and graph optimization algorithms.
They should hold a Master’s degree, preferably in operations research or applied mathematics, and have solid programming skills in C++ and/or Python

Applications must be sent to the three FEMTO-ST supervisors listed below. The
application should include at least:
- A detailed CV
- A copy of an identity card or passport
-Academic transcripts from the first and second years of the Master’s program (or most recent degree)
- A motivation letter explaining the candidate’s interest in pursuing a PhD and conducting research on the proposed topic
Application deadline : 15 mai 2025
Start of contract : October 2025

https://www.femto-st.fr/en/job-training-thesis/phd#job-5761

Applied Mechanics: "Solid hydrogen storage by reversible hydriding"

The thesis will be housed in the Applied Mechanics Department (~110 people including ~50 teacher-researchers, researchers and technicians). The Mat'éco team is structured around two areas of application: bio-based composite materials and hydrogen storage. It has around ten PhD and Masters students working on multiscale and multimodal experimental techniques and numerical simulations using finite elements or discrete elements, with the aim of considering thermo-mechanical couplings as well as chemical, hydrous or gaseous couplings.

Context :
The relevance of hydrogen as an energy carrier and its role in the ecological transition are well established. Safe, energy-efficient storage solutions are the sine qua non of its green credentials. A long-term storage solution will need to be able to reinvest green hydrogen from electrolysis during peaks in consumption, thanks to the overproduction of electricity from renewable sources (wind, hydro, solar, etc.) during periods of under-consumption. The solid storage of H2 by reversible hydriding of intermetallic alloys, usually in powder form, is a promising technology because it is safe, requires no heavy infrastructure and offers excellent energy efficiency, since it is produced at low pressure (from 1 to 10 bars: little compression, fewer leaks) and at near-ambient temperatures (between 10 and 80°C).

DESCRIPTION OF CYCLAHMID PROJECT (CYCLage des Alliages Hydrurés : MIcrostructure et Décrépitation) :
This thesis is part of a Graduate School EUR EIPHI project, with support from the Bourgogne-Franche Comté region. The CYCLAHMID project explores the problem of the microstructure - storage - decay link through the comparative study of two intermetallics and an HEA. In this project, you will benefit from the combination of DMA/FEMTO ST's expertise in multi-scale mechanical characterisation and working under hydrogen and PMDM/ICB's expertise in structural and microstructural characterisation of high-reactivity materials.

PROFILE REQUIREMENTS :
Master student in Mechanical Engineering or Materials Sciences, final year of engineering school or already graduated.
• Knowledge in mechanics of metallic materials, including production processes, characterization and mechanical and microstructural modelling, are expected. An interest in experimentation, materials characterization and the development of new experimental techniques would be an advantage. Knowledge of solid phase change materials or in hydrogen as an energy carrier frame would be appreciated, but is not essential.
• Most of all, we are looking for candidates who are rigorous, methodical and committed.
• Fluency in written and spoken English is essential for scientific communication (articles, international conferences).

Starting date : 2025 September / October

https://www.femto-st.fr/en/job-training-thesis/phd#job-5720

AS2M : "Estimation and control of flexible structures with physics informed learning"

Context : This thesis will mainly take place within the ROMOCO team at AS2M (Automatique et Systèmes Micro-Mécatroniques) department of FEMTO-ST Besançon. The aim of this thesis is to develop identification, estimation and control methods for flexible structures (1D and 2D) actuated by HASEL actuators, combining artificial intelligence (AI) techniques and the port Hamiltonian framework.

This thesis aims to address the estimation and control problem of HASEL actuator driven flexible structures (1D and 2D)(experimental set-up in 1D case existing in Department AS2M of FEMTO-ST institute  using the port Hamiltonian structure informed Neural network.

Candidates profile :
• Excellent MSc/Engineer in Automatic Control, Applied mathematics, Robotics.
• Strong knowledge background in automatic control and/or applied mathematics, experience in
Artificial Intelligence.
• Fluent in speaking and reading English.

Funding and application :
The Ph. D thesis may start in October 2025.

Please send your application documents, including a detailed CV, a cover letter, all transcripts, and recommendation letters to both advisers, Prof. Yann Le Gorrec (legorrec@femto-st.fr), Dr. Yongxin Wu (yongxin.wu@femto-st.fr

https://www.femto-st.fr/en/job-training-thesis/phd#job-5713