Energy-based analog inverse solvers for frugal AI in nondestructive evaluation M/F

Job Description

Thesis Subject

What if some inverse problems in imaging could be solved not by power-hungry iterative code, but by the natural relaxation of a reconfigurable analog circuit? This PhD sits at the interface of frugal AI, signal processing, hardware and nondestructive evaluation.

Within the ENACT AI Cluster, you will develop energy-based analog inverse solvers for nondestructive evaluation (NDE) on Field-Programmable Analog Arrays (FPAAs) — the analog counterpart of FPGAs. The scientific ambition is twofold: push physically grounded inverse methods beyond standard digital pipelines, and rigorously quantify reductions in energy per solve compared with optimized CPU/GPU and FPGA baselines.

The project targets three real THz-NDE use cases: (i) sparse deconvolution of THz impulse responses to localize subsurface interfaces beyond nominal axial resolution; (ii) multilayer stratigraphic reconstruction under physical constraints such as ordering and minimum thickness; and (iii) localized coating-thickness estimation on regions of interest for scalable inspection. You will derive the energy functions, implement analog relaxation dynamics on FPAA hardware, and build the hybrid analog-digital loop for calibration, orchestration, convergence detection, and uncertainty/quality indicators.

This PhD offers a rare environment: cotutelle with Georgia Tech, strong interaction with Institut Jean Lamour for FPGA/digital acceleration, and direct access to the ANR AATLAS platform. You will work on real experimental data, reconfigurable hardware, and scientifically ambitious low-footprint AI.

Your Work Environment

The doctoral candidate will be recruited by CNRS within the Georgia Tech–CNRS International Research Lab (IRL 2958), hosted at Georgia Tech Europe in Metz, in the THz Imaging & Nonlinear Dynamics team led scientifically by Alexandre Locquet. Contract: 36-month full-time CNRS doctoral contract. Expected start: 1 October 2026. PhD enrolment: Université de Lorraine (doctoral school C2MP).

The project is part of the ENACT AI Cluster, in the priority area “AI for Engineering and Scientific Discovery”, with a strong scientific and environmental ambition: developing lower-footprint AI methods for real inverse problems in nondestructive evaluation. The topic has already passed the first ENACT selection stage; the current objective is to identify an outstanding candidate quickly for the final interview phase. This PhD offer is provided by the ENACT AI Cluster and its partners. Find all ENACT PhD offers and
actions on https://cluster-ia-enact.ai/.

You will work in a rare environment at the intersection of frugal AI, analog computing, reconfigurable electronics and THz imaging. The PhD is directly connected to the ANR PRC AATLAS project (coordinated by Alexandre Locquet), which provides the FPAA+FPGA platform, hardware ecosystem and energy-benchmarking framework. The thesis will be carried out in cotutelle with the Georgia Institute of Technology (Atlanta, USA) under the supervision of Prof. Jennifer Hasler, a pioneer of FPAA technology, with French co-supervision by Prof. Slavisa Jovanovic (Institut Jean Lamour) for the FPGA/hybrid-architecture aspects. Extended stays in Atlanta are planned.

This position is especially attractive for a candidate who wants both high-level science and hands-on prototyping: modelling, coding, circuits, measurements and validation on real data. Applications will be reviewed on a rolling basis to match the ENACT timeline, so early applications are strongly encouraged. Please include a detailed CV, motivation letter, Bachelor's/Master's transcripts (or equivalent), and, if possible, contact details for one or two academic referees. Scientific contact: alocquet@georgiatech-metz.fr

• Degree: MSc/engineering degree (or equivalent) in electrical engineering, applied physics, computer science, applied mathematics, signal processing, or a related field.
• Strongly expected skills: Python programming; solid background in optimization and/or inverse problems; strong interest in signal processing, experimental data and physics-based modelling.
• Important assets: experience with electronics, FPGA/embedded systems, analog circuits or mixed-signal systems; familiarity with NumPy/SciPy/PyTorch; interest in hardware-algorithm co-design and energy-efficiency evaluation.
• Personal qualities: curiosity, autonomy, scientific rigour, willingness to build and test prototypes, and ability to thrive in an international interdisciplinary environment. Fluency in English is essential; extended stays in Atlanta are planned within the cotutelle.

Compensation and benefits

Compensation

2300 € gross monthly

Annual leave and RTT

44 jours

Remote Working practice and compensation

Pratique et indemnisation du TT

Transport

Prise en charge à 75% du coût et forfait mobilité durable jusqu’à 300€

About the offer

About the CNRS

The CNRS is a major player in fundamental research on a global scale. The CNRS is the only French organization active in all scientific fields. Its unique position as a multi-specialist allows it to bring together different disciplines to address the most important challenges of the contemporary world, in connection with the actors of change.

CNRS

The research professions