38 Postdoctoral Fellowship at Inria, France

Atlantis is  a joint project-team  between Inria and  the Jean-Alexandre Dieudonné Mathematics Laboratory at  Université Côte d’Azur. The team  gathers applied mathematicians and  computational scientists who are collaboratively undertaking  research activities aiming at the design, analysis, development and  application of advanced numerical methods for solving systems of  partial differential equations (PDEs) modelling nanoscale light-matter interaction problems. In this context, the team is  developing  the   DIOGENeS  [https://diogenes.inria.fr/]  software suite,  which  implements  several Discontinuous  Galerkin  (DG)  type methods tailored to the systems  of time- and frequency-domain Maxwell equations  possibly coupled  to  differential  equations modeling  the behaviour of propagation  media at optical frequencies. DIOGENeS also includes a component dedicated to the optimization of geometrical characteristics of nanostructures driven by some performance objective in the contex of inverse design strategies of nanophotonic setups.   DIOGENeS is a unique  numerical   framework  leveraging   the  capabilities   of  DG techniques  for  the simulation  of  multiscale  problems relevant  to nanophotonics and nanoplasmonics.

Candidates for postdoctoral positions are recruited after the end of their Ph.D. or after a first post-doctoral period: for the candidates who obtained their PhD in the Northern hemisphere, the date of the Ph.D. defense shall be later than September 1, 2022; in the Southern hemisphere, later than April 1, 2022. In order to encourage mobility, the postdoctoral position must take place in a scientific environment that is truly different from the one of the Ph.D. (and, if applicable, from the position held since the Ph.D.); particular attention is thus paid to French or international candidates who obtained their doctorate abroad.

The postdoc focuses on novel and advanced embedded AI, combining two complementary aspects. On the one hand, efficient data communication using AI [2,3] with neural network models that can summarize and compress data from one or more sensors to a server. On the other hand, TinyML and TinyMLops [1] which focus on the implementation of AI directly on constrained microcontrollers.

TinyML aspect: The goal is to implement efficient AI model execution (TinyML) on microcontrollers, and manage AI models (MLOps: remote updates, performance monitoring – here secure TinyMLOps) on hardware such as Nordic nRF52, STM32, ESP32, or RISC-V. Networking will use IoT technologies such as BLE, 802.15.4, or LTE-M.  
On top of this hardware, prototypes will be developed in conjunction with an open-source operating system written in embedded Rust (Ariel OS [4]) or embedded C (RIOT [5]).  
These prototypes will be co-developed and tested with Freie Universität Berlin. This project follows up on RIOT-ML (see below [6]), also linked to concrete industrial use cases for efficient sensor-to-server communication (Digital Twins).

The Internet was conceived to enable computer resources’ time-sharing, but soon its main function became to deliver content to end users, but it is now called to play a new key role: to pervasively support machine learning (ML) operation both for model training and prediction serving.

There are two aspects calling for Internet-wide deployment of ML systems. First, data—one key ingredient of ML success—is often generated by users and devices at the edge of the network. The classic ML operation in the cloud requires such data to be collected at a single computing facility where training occurs. Data aggregation can be very costly, or simply impossible because of capacity constraints, privacy issues, or ownership ones. These scenarios call for distributed learning systems, where computation moves, at least in part, to the data. For example, Google’s federated learning [mcmahan17,kairouz21] enables mobile phones, or other devices with limited computing capabilities, to collaboratively learn an ML model while keeping all training data locally. Distributed ML training is already a difficult task in a cluster setting. Indeed, optimization techniques, distributed systems, and ML models are a triad difficult to untangle: e.g., relaxed state consistency across computing nodes increases system throughput but may jeopardize convergence of the optimization algorithm or affect the final solution selected, leading to models with very different generalization capabilities [chen16]. Additional challenges arise when training moves to the Internet. First, the system potentially scales up to billions of devices, against at most thousands of GPUs to break ML training records in a cluster. Second, local datasets are highly heterogeneous with very different sizes and feature/label distributions. Third, devices may have very different hardware and connectivity. Fourth, communications are often unreliable (devices can be switched off at any time), slow (latencies are 2 orders of magnitude larger), and expensive for battery-constrained devices. Fifth, privacy concerns are often important and limit the operations that can be performed during training to avoid inadvertently disclosing sensible information. Finally, training is more vulnerable to malicious attacks. For all these reasons, federated learning (as ML training over the Internet is now usually called) has emerged in the last years as a specific research topic—well distinct for example from high-performance computing or cloud computing—at the intersection of machine learning, optimization, distributed systems, and networking.

The goal of the project-team is to better understand and formalize the effects of complex environments on the dynamics of the interconnections, as well as to develop methods and tools for the analysis and control of such systems. 1) Modeling of complex environments : The environment is seen as a dynamical object in order to model phenomena such as a temporary loss of connection, a nonhomogeneous environment or the presence of the human factor in the control loop but also the problems involved with technological constraints. 2) Questions of stability characterization or robust stabilization of possibly nonlinear infinite-diemensional are considered by various methods : $H_\infty$-control, nonlinear control via Lyapunov-Krassovski techniques, observers, adaptative control, predictive control, set invariance. 3) Determination of finite-dimensional controllers of low order for infinite-dimensional systems.

The recruitee will work in close relation with both teams, Scool and the Inserm team.

With Scool, the goal is to explore how the problem at hand can be modeled, most likely as a contextual bandit problem, propose algorithms, study their theoretical properties.

With the Inserm team, the goal is to investigate the application of this work to the real practical case.

These two lines of work are intrinsically strongly tied to each others.

The recruited post-doc will collaborate with colleagues in the MAGNET team and the TRUMPET, FLUTE and REDEEM project consortia.

If the research features a prototype, it will contribute to the project’s open source library.

We hope the post-doc can bring new expertise to the group and/or can help intensifying collaboration in the project consortium.  He will collaborate closely with the other group members on realizing the research objectives of the project.  Engineers in the team can support the prototyping and validation.

Possible topics of research include (but are not limited to):

• Cryptography-based strategies to improve the security of privacy-preserving AI systems.
• Inference methods for privacy assessment and/or theory for statistical privacy in general
• Programming language strategies such as those relevant in compilers and interpreters
• Design and development of the TRUMPET/FLUTE platform and its supporting algorithms.

The research project will explore quantum protocols based on the concept of quantum nonlocality and quantum networks (see arXiv:2104.10700). A non-exhaustive list of potential projects is:

• Methods for characterizing quantum correlations beyond the Bell scenario:
  • mathematical foundation of these methods (C* algebras, noncommutative polynomial optimization): see e.g. arXiv:2210.09065, arXiv:2212.11299, arXiv:2301.12513
  • improve/find new algorithms for characterizing these correlations
  • numerical developpement of these algorithms, see e.g. arXiv:2211.04483
• Understanding the foundational implications of quantum correlations in networks, see e.g. arXiv:2101.10873 and arXiv:2105.09381
• Develop the applications of network nonlocality to certification protocols, such as
  • randomness generation: arXiv:2209.09921 
  • self testing of measurements and states: arXiv:1807.04956, arXiv:2201.05032
• Adapt existing protocols for their experimental implementation
• Develop practical benchmarks of the concept of ‘Genuine Multipartite Nonlocality’ introduced in arXiv:2105.09381
• Develop SDP relaxations for condensed matter problems, see e.g. arXiv:2212.03014, arXiv:2310.05844, arXiv:2311.18707, arXiv:2311.18706 
• Explore the limits of quantum distributed computing, see e.g. arXiv:1810.10838, arXiv:0903.113

Any other suggested research project in quantum information theory can be discussed (both from the physics, the computer science or the mathematical viewpoint).

The recruited candidate will be responsible for designing and developing an LLM-based approach to query the MetaKH knowledge graph, providing personalized support as well as visual and textual solutions tailored to user needs. They will also be tasked with designing, implementing, and deploying evaluation methods with users to assess the usefulness and reliability of the assistant’s results, thereby contributing to the improvement of decision-making processes in metabolomics.

The expected deliverables are:

• Publication: An approach leveraging LLMs to replicate the methods of research assistants.
• Publication: A method for evaluating conversational interfaces based on LLMs.
• Software: A stable version of the AI assistant’s source code and the web application.

The recruited candidate will collaborate with Aline MENIN and Marco WINCKLER on aspects related to human-computer interaction and data visualization, as well as with other researchers of Wimmics for topics related to the Semantic Web and LLMs.

The Post-Doc chosen for this project should have a broad range of analytic skills including at least several of the following:

Developing experimental methodologies that allow the rigorous collection of hyperscanning data.

Managing hyperscanning experiments, including the use of PsychoPy software to develop triggers to allow alignment of the different kinds of data (language, nonverbal behavior, inter-brain synchrony).

Knowledge of hyperscanning technologies, particularly functional Near-InfraRed Spectroscopy (fNIRS). Familiarity with the NIRx system is a plus.

Analysis of hyperscanning neuroimaging data, using advanced statistics and machine learning methodologies for temporally-sensitive data, such as GLMM, Random Forests, LSTM, etc..

Experience in programming in Python and use of relevant architectures, software, and libraries, such as HyPyP.

The successful candidate will work in a multi-disciplinary team to conduct hyperscanning experiments with adults and children, and will be responsible for analyzing the resultant data, and for collaborating to write-up and publish the results.

The proposed position will focus on the development of grant-free access techniques for IoT applications.

One common communication scenario in IoT applications is massive machine-type communications (mMTC), where a large number of devices transmit sporadic, small packets. In traditional cellular systems, each device is allocated orthogonal resources prior to uplink transmission via a grant mechanism. However, this allocation requires signaling on control channels, which can exceed the data payload size and lead to inefficient resource use. Consequently, grant-free methods [1] that eliminate or reduce control traffic are well suited for these scenarios. Removing coordination introduces non-orthogonality, resulting in the superposition of signals from some or all devices.

A recent family of random access protocols—sometimes called “modern random access”—aims to address and even exploit this phenomenon. The IRSA protocols (Irregular Repetition Slotted ALOHA) [2,3,4] use Successive Interference Cancellation (SIC) and represent one form of grant-free technique, but they can also operate with any packet transmission scheme. Adapting these methods for grant-free mMTC in cellular networks is therefore of prime interest and the main objective of this position.

The Agile Language Engineering (ALE) associate team represents a long and successful collaboration between Inria and CWI. Its focus is on improving software engineering through language engineering: both for the end-users and the language engineers themselves. One main topic in the team is live programming: programming while the application runs, without intermittent shutdown and restarts. A language engineering perspective on live programming requires rethinking the way languages are implemented. For this we propose the concept of LipVM: the Live and Interactive Programming Virtual Machine. Targeting this VM will allow language engineers to abstract from the technical challenging aspects of live programming, such as hot-swapping, state migration, back-in-time debugging, and orthogonal persistence.

The post-doc will contribute to developing this concept, its design, and its prototype implementation, performing research at the cutting edge of virtual machine engineering, language engineering, and programming language design, in close collaboration with Inria DiverSE and the Software Analysis & Transformation (SWAT) group at CWI. We expect that LipVM will be a stepping stone to bring live programming to programming languages, modeling languages, and domain-specific languages. The postdoc position will further strengthen the collaboration between Inria and CWI.

(18) Postdoctoral Fellowship Position

Postdoc Fellowship Position summary/title: Post-Doctoral Research Visit F/M Avatar embodiment and multisensory integration in immersive populated spaces

Deadline : 2025-06-17

View details & Apply

Candidates for postdoctoral positions are recruited after the end of their Ph.D. or after a first post-doctoral period: for the candidates who obtained their PhD in the Northern hemisphere, the date of the Ph.D. defense shall be later than September 1, 2022; in the Southern hemisphere, later than April 1, 2022.

In order to encourage mobility, the postdoctoral position must take place in a scientific environment that is truly different from the one of the Ph.D. (and, if applicable, from the position held since the Ph.D.); particular attention is thus paid to French or international candidates who obtained their doctorate abroad.

Candidates for postdoctoral positions are recruited after the end of their Ph.D. or after a first post-doctoral period: for the candidates who obtained their PhD in the Northern hemisphere, the date of the Ph.D. defense shall be later than September 1, 2022; in the Southern hemisphere, later than April 1, 2022.

In order to encourage mobility, the postdoctoral position must take place in a scientific environment that is truly different from the one of the Ph.D. (and, if applicable, from the position held since the Ph.D.); particular attention is thus paid to French or international candidates who obtained their doctorate abroad.

We seek a postdoctoral researcher interested in exploring adaptive input pipelines for software and systems used in digital fabrication. This includes real-time sensing and processing of user input in tools that support:

• Interactive fabrication or semi-automated fabrication
• CAD and 3D modeling, including AI-assisted modeling and generative design
• Augmented reality modeling and in-situ model visualization

Your work will contribute to re-design how such tools capture, interpret, and respond to user input to support precision, expressivity, and collaboration. Your work will expand insights into input pipelines and refine technologies that are part of the INPUT project

The recruited person, supervised by Elaine Pimentel (UCL) and Alexis Saurin (Picube team) is expected to conduct research on proof representations, using linear logic and fixed-point logic to design decompositions of logical modalities; this research will span :

1) from pure proof-theoretical research, aiming at understanding how linear logic an the mu-calculus can be used to decompose a variety of modal logics, aiming at providing a uniform account of those various logics, factoring structuring theorems through the properties of the encodings and providing a study of ecumenical proof systems with fixed-points;

2) to computational interpretation of logics, by developing cut-reduction systems for various systems adapted from Kozen’s mu-calculus;

3) via the mechanization of the meta-theory of the above proof-theory in the Coq/Rocq proof assistant. 

The goal of this project is to address key issues in Large Language Models (LLMs), particularly cultural biases stemming from Western-centric training data. These models often underperform or exhibit prejudice in non-English and especially South American contexts due to limited resources for bias detection. We propose to define sociologically grounded notions of social bias that can be computationally identified and measured. This involves data collection, annotation, and adapting existing datasets. Once biases are defined, we will detect them through model behavior analysis and apply cutting-edge fact-editing techniques to adjust the model’s internal weights, mitigating harmful stereotypes while enhancing culturally relevant knowledge. Our focus is on general, multilingual methods, with a key application to Latin American languages and cultural contexts.

The postdoc is funded by the project PDE-AI of PEPR IA. The recruited researcher will work with people from McTAO team in Sophia Antipolis and LJAD CNRS math lab in Nice.

The aim of this 2-year mission – co-funded by Airbus and Inria – is to work on low-rank compression algorithms, which are a fundamental building block of the H-matrix approach. There is a huge variety of algo’s fulfilling this role, between classical approaches (SVD, ACA, CUR, ID, …), randomized variants, reconciliations between these methods and other usual algo’s (ACA is for instance a way to rewrite LU), … Today, H-matrix contains only a few of these algorithms. The aim of the post-doc is :

• enrich the available choice of low-rank compression algo ;
• implement them in a composable way, so that they can be used in the h-matrice library as well as in any other software library;
• to know for each algo its range of effectiveness, in terms of input data (size), target machine (CPU, GPU), precision, execution time, etc.
• be able to dynamically select the most suitable algo at runtime.

As the H-matrices tools at Airbus and Concace are halfway between industry and academic research, the aim of the research within this mission is to identify the most efficient numerical techniques and pass them on to industry for implementation in design offices.

Crop protection often remains dependent on chemical pesticides, which are both harmful for the environment and human health. Resistant crops are an agroecological alternative to pesticides, but their extensive use may lead to the emergence/selection of virulent pathogens and resistance breakdown. Devising deployment strategies of resistant crops that are both efficient, i.e. that reduce crop damages, and durable, i.e. that limit the virulent pathogen populations, is hence a major issue.

The postdoctoral fellow will tackle this issue by means of a demo-genetic model, tailored for a specific pathosystem, the phoma stem canker of oilseed rape caused by fungus Leptosphaeria maculans. The emergence and development of virulent pathogens may vary according to the genetic determinisms of virulence (molecular mechanisms responsible for the transition to virulence, epistatic interactions, fitness costs), which are studied by other partners of the ENDURANCE project.

Since the late 19th century, microbiologists have been developing culture media and techniques to study microorganisms such as bacteria, archaea, yeasts, and filamentous fungi. However, only a small fraction of bacterial species are currently cultivable, severely limiting their study. While advances in metagenomics have highlighted this limitation, genome sequencing alone has not provided sufficient explanations for why many bacterial species remain unculturable. This lack of cultivability remains a scientific enigma and a major barrier to applications in fields such as disease control, antibiotic discovery, biomolecule production, microbiome management, and bioremediation. In bacteria that live in association with eukaryotic hosts, understanding metabolic interactions with their partners opens new possibilities for developing tailored culture media, thereby facilitating their study. Using grapevine Flavescence Dorée phytoplasma as a proof of concept, the TARGET project aims to demonstrate that integrating systems biology, genome engineering, and culturomics can help overcome the challenge of bacterial non-cultivability.

This postdoctoral position is part of the project’s metabolic modelling work package, with objectives including the development of numerical models for the metabolism of minimal bacteria and Flavescence Dorée phytoplasma. These models will be used to predict suitable culture media. The project involves close collaboration between modellers and microbiologists to bridge computational and experimental approaches.

The recruited researcher will work in collaboration between Inria Pleiade team in Talence and INRAE BFP (Fruit biology and Pathology) unit at Villenave d’Ornon. At Inria, the work will be carried under the supervision of Clémence Frioux. Pleiade hosts researchers in mathematics, computer science and computational biology working on developping approaches for microbioal community characterisation. Two teams are involved at INRAE: MOLLI (Pascal Sirand Pugnet) where most of the experimentations will be carried out, and META specialised in metabolic modelling. The co-supervisor of the position is Sylvain Prigent, researcher in the META team.

Ruminants play a significant role in human nutrition and food security. Ruminants can harvest nutrients from forage diets rich in fibres and transform them into human-edible products with high-quality proteins. Feed transformation occurs mainly in the rumen through a metabolic cascade of hydrolytic and fermentative reactions carried out by a complex microbial community (rumen microbiota) constituted by hundreds of species that include bacteria, archaea and eukaryotes. During rumen fermentation, methane (CH4) is produced and eructed by the animal. This CH4 contributes to 44% of greenhouse gas emissions from the livestock sector.

Reducing methane (CH₄) emissions from ruminants is a major challenge for the livestock sector. An optimal CH₄ mitigation strategy should also induce co-benefits such as enhanced animal productivity and health. CH₄ is produced during the fermentation of feeds in the rumen. This process is carried out by a complex microbial community (rumen microbiota) and mediated by hydrogen (H₂) in the rumen ecosystem. Our knowledge on the drivers that shape H₂ flows is still incomplete. H₂Rumen aims to generate fundamental knowledge on H₂ flows in the rumen ecosystem and to translate this knowledge into predictive mathematical models of rumen fermentation. Our scientific outputs will be of high value for the optimal design of CH₄ mitigation strategies with co-benefit for the animal. Our hypothesis is that thermodynamics and microbial interactions jointly control hydrogen transactions in the rumen ecosystem. H2Rumen addresses the following fundamental scientific question: Where does H₂ go? We will answer this question with an integrative approach that combines in vitro experiments, in silico modelling and omics approaches. In addition to the impact on ruminant livestock, our methods might be applicable to other ecosystems such as the human gut, engineering bioreactors and fermented food ecosystems.

In this context, the goal of the post-doc is thus to help characterizing different possible side-effects of radiotherapy and in particular white matter lesions, identified as white matter hyperintensities (WMH) in the patient brain. We will aim at investigating and developing advanced spatio-temporal (ST) models and AI tools [Young et al 2024, Couronne et al 2019] to extract, if it exists, a set of ST features which characterize WMH of different nature that may be associated either to post-RT side-effects (eg radio-induced leukoencephalopathies, radio-necrosis, post-RT oedema) or to treatment responses (eg  brain tumor progression, peritumoral oedema). So doing we will help in improving knowledge about the radiosensitivity of healthy brain tissues.

An increasing number of applications rely on complex inference tasks based on machine learning (ML). Currently, two options exist to run such tasks: either served directly by the end device (e.g., smartphones, IoT equipment, smart vehicles) or offloaded to a remote cloud. Both options may be unsatisfactory for many applications: local models may have inadequate accuracy, while the cloud may fail to meet delay constraints. In [SSCN+24], researchers from the Inria NEO and Nokia AIRL teams presented the novel idea of inference delivery networks (IDNs), networks of computing nodes that coordinate to satisfy ML inference requests achieving the best trade-off between latency and accuracy. IDNs bridge the dichotomy between device and cloud execution by integrating inference delivery at the various tiers of the infrastructure continuum (access, edge, regional data center, cloud). Nodes with heterogeneous capabilities can store a set of monolithic machine-learning models with different computational/memory requirements and different accuracy and inference requests that can be forwarded to other nodes if the local answer is not considered accurate enough.

Unlike the traditional finite element method, which relies on an explicitly given approximation space (typically piecewise polynomial), in multiscale numerical methods the approximation space is driven numerically by the PDE model, incorporating fine-scale details of the domain geometry and coefficient distribution. The multiscale methods developed by the associate team can be interpreted as approximate substructuring techniques, where the interiors of macro-cells are eliminated through a low-dimensional parametrization of either Neumann data (as in MHM [1,3]) or Dirichlet data (as in Trefftz methods [2]). Since the computation of the approximation basis is local to the coarse cells, multiscale numerical methods are highly parallelizable, which allows them to benefit from increasing computational facilities while keeping communications very low. Alternatively, multiscale basis functions can be “learned” using machine learning techniques, which makes multiscale methods even more accessible.

The research program of this postdoctoral program will focus on error analysis of multiscale numerical discretization methods for nonlinear problems, and their integration with domain decomposition and scientific machine learning approaches.

[1] Araya, R., Harder, C., Paredes, D., & Valentin, F. (2013). Multiscale hybrid-mixed method. SIAM Journal on Numerical Analysis, 51(6), 3505-3531.

[2] Boutilier, M., Brenner, K., & Dolean, V. (2024). Robust methods for multiscale coarse approximations of diffusion models in perforated domains. Applied Numerical Mathematics, 201, 561-578.

[3] Gomes, A. T. A., Pereira, W. S., & Valentin, F. (2023). The MHM method for linear elasticity on polytopal meshes. IMA Journal of Numerical Analysis, 43(4), 2265-2298. 

As technology advances, we are seeing an increasing number of objects communicating via more and more different communication technologies. These technologies enable objects to self-organise to form a network that evolves with their mobility, or to connect to a communication infrastructure such as a cellular or wifi network, for example.

The cohabitation of these two types of communication (node-to-node or via an infrastructure) can be useful in places where the infrastructure is unavailable or unreliable, as one mode can take over from another depending on the situation. Decisions on whether or not to transmit information are taken by so-called opportunistic routing protocols.

The literature provides some interesting elements in the field of opportunistic routing. Nevertheless, they have certain limitations that can be partially overcome by introducing the consideration of node mobility types into information transfer decisions. Our previous work has shown that a node is able to determine its mobility type by observing the different networks in its environment.

The main objective of this project is to study the best strategy based on this knowledge of mobility to achieve efficient data transmission in mobile scenarios. 

This work will be carried out jointly by Inria’s FUN and COATI teams.

With the help of Victor David and Anthony Hunter, the recruited person will build upon existing symbolic and formal models of implicit arguments to design learning-based methods and evaluation metrics. The objective is not to propose new symbolic models, but rather to translate and operationalize these theoretical insights into computational approaches for explanation generation and evaluation. Victor and Anthony will provide close guidance to ensure the proper transfer and integration of formal knowledge.

In parallel, the postdoctoral researcher will work in close collaboration with Serena Villata and Elena Cabrio, experts in natural language processing (NLP) and argument mining, who will supervise the NLP components, dataset exploration, and empirical evaluation protocols. This interdisciplinary framework ensures that both formal rigor and empirical relevance are at the core of the project.

Candidates for postdoctoral positions are recruited  after the end of their Ph.D. or after a first post-doctoral period: for the candidates who obtained their  PhD in the Northern hemisphere, the date of the Ph.D. defense shall be later than September 1, 2022; in the Southern hemisphere, later than April 1, 2022.

In order to encourage mobility, the postdoctoral position must take place in a scientific environment that is truly different from the one of the Ph.D. (and, if applicable, from the position held since the Ph.D.); particular attention is thus paid to French or international candidates who obtained their doctorate abroad.