Internship offres

05 déc., 2025

STAGE, Stage

CEA 48.7126483,2.1934657

Although formal verification is essential for ensuring the safety and security of software, it remains difficult to deploy and use effectively by non-experts due to its steep learning curve. Recent advances in large language models (LLMs) have demonstrated remarkable abilities in code understanding, synthesis, and reasoning. These advances open promising research directions for assisting developers and verification engineers in formal specification and verification tasks. The goal of this internship is to explore and evaluate the integration of LLMs assistance into the Frama-C environment to support and automate parts of the specification and verification workflow. The work will focus on identifying the extent to which LLMs can provide meaningful support without compromising the rigor and reliability of formal program analysis. The following topics represent potential research and technical directions of the internship. Depending on the interests of the intern, one or more of them...

RequiredInterest in AI-assisted software engineering Willingness to explore interactions between LLMs and formal verification frameworks Solid knowledge of Python and its ecosystem Ability to work in a team PreferredFamiliarity with machine learning and large language models, including prompt design or API integration. Familiarity with the Frama-C platform. Some knowledge of the OCaml and C programming languages

15 nov., 2025

STAGE, Stage

Internship - Secret Quorums Protecting Byzantine Reliable Broadcast Against Adaptive Adversaries H/F

CEA 48.7126483,2.1934657

Context: Modern distributed systems, such as blockchain consensus and secure multiparty computation protocols, relies on a fundamental primitive: byzantine reliable broadcast. It allows a set of processes to agree on a message broadcasted by a dedicated process, even when some of them are malicious (byzantine). It relies on quorum systems to prevent correct processes from delivering two different messages: each message needs to be certified by a majority (a quorum) of processes before delivery. Byzantine reliable broadcast needs to maintain low latency and small communication complexity, but its efficiency depends on the number of processes constituting a majority. It was proposed in [1] to validate each message by smaller quorums such that 2 quorums for 2 different messages will intersect only with high probability. This technique is not yet secure against fully rushing-adaptive adversaries: a malicious entity can learn the processes constituting a small quorum and corrupt them...

Methodology: The intern will have the following responsibilities: Prepare a state-of-the-art on similar solutions (this would be part of the state of art section for a research paper). Become familiar with protocols already proposed in [1] and [2]. Specify a new distributed protocol solving the problem and proving its correctness. Code a prototype and evaluate its practical relevance. Prepare a research paper. Competences: Being Master 1 or 2 in Computer Science/Engineering. Knowledge about distributed systems in general. Have an interest in theoretical computer science. Basic Rust abilities.

23 oct., 2025

STAGE, Stage

Internship - Develop a 3D Multi-Modal Annotation Tool H/F

CEA 48.7126483,2.1934657

Annotationg in 3D such (detection, segmentation ), is required to train reliable perception models, which can be used in autonomous driving for example. However such a process requires multi-modal (camera, LIDAR, RADAR) visualization and annotation which is a complicated and costly process. This is why at CEA-List we aim to develop smart tools to facilitate this annotation process. The objective of the internship is to allow for multi-modal annotation and visualization of 3D scenes. For example, annotate a scene observed by an autonomous vehicle with both cameras and LIDAR sensors. Develop both backend and front end for our open source annotation tool PIXANO to allow for : Multimodal visualization (camera / LIDAR / bev / 3D) Visualization of Nerf and/or Gaussian splatting Annotation in 2.5D at bird's eye view

Students in their 4th or 5th year of studies (M1, M2 or gap year) Computer vision skills Strong interest for 3D Coding skills relevant for both frontend and backend development is best (Python, JavaScritp, SQL)

15 oct., 2025

STAGE, Stage

Final Year Internship Modeling The Scattering Of Elastic Waves By Flaws H/F

CEA 48.7295843,2.1483258

Simulation of ultrasonic Non Destructive Testing (NDT) is helpful for evaluating performances of inspection techniques and requires the modeling of waves scattered by defects. Two classical flaw scattering models have been previously usually employed and evaluated to deal with inspection of planar or multifaceted defects, the Kirchhoff approximation (KA) for simulating reflection and the Geometrical Theory of Diffraction (GTD) for simulating diffraction. The student will theoretically study existing modified versions of the two previous approximations to deal with rough defects. These two methods will then be implemented, numerically evaluated, and compared with each other and with a reference numerical method. Conformément aux engagements pris par le CEA en faveur de l'intégration des personnes handicapées, cet emploi est ouvert à toutes et à tous. Le CEA propose des aménagements et/ou des possibilités d'organisation pour l'inclusion des travailleurs handicapés.

Fin études ingénieur / Master 2 ; mécanique /acoustique /mathématiques appliquées.

05 déc., 2025

STAGE, Stage

Stage en Cryptologie H/F

CEA 48.7126483,2.1934657

Transciphering is a cryptographic technique that re-encrypts data from one scheme to another without an intermediate decryption step. This process drastically reduces the overhead induced by the size of homomorphic ciphertexts during their transmission and storage. The core idea of transciphering is to convert data encrypted with a classical symmetric encryption scheme into a Homomorphic Encryption (HE) scheme. How Transciphering Works ? Consider a scenario with a plaintext message m, a symmetric scheme SYM with key k, and a Homomorphic Encryption scheme HE with a public key pk. The message is initially encrypted as SYM.Enc_k(m). Transciphering uses the evaluation capability of the HE scheme to perform the symmetric decryption process entirely within the homomorphic domain. The process requires the symmetric key k to be encrypted under HE public key, HE.Enc_pk(k). By using the homomorphic evaluation key evk, the result is a message encrypted directly under HE public key pk:...

Durée (Mois):

6

01 déc., 2025

STAGE, Stage

Accurate 3D Scene Reconstruction From Images With Neural Method H/F

CEA Île-de-France

Missions: During this internship, you will explore the state of the art techniques for 3D scene reconstruction from 2D image using Neural Fields. With a focus on the accuracy of reconstructed geometry, you will design and develop a method of 3D surface reconstruction, which will be applied to industrial environment. The internship will notably include: Reviewing existing literature on the topic Selecting and implementing an existing approaches in the Neural Fields framework NeRFStudio Benchmarking the new method using images capturing an industrial environment Job-related benefits Joining the CEA List and the LVML as an intern means: Working in one of the most innovative research organizations in the world, addressing societal challenges to build the world of tomorrow Discovering a rich ecosystem: privileged connections between the industrial and academic sectors Conducting research autonomously and creatively: encouragement to publish results (scientific articles, patents,...

Qualifications: Students in their 5th year of studies ( Master 2) or gap year Computer vision skills (3D vision, image processing) Python proficiency in a deep learning framework (preferably PyTorch

20 nov., 2025

STAGE, Stage

Formal Methodology For The Exploration And The Evaluation Of Complex Critical Sw Architecture M - F H/F

CEA 48.735923981,2.166033197

The internship aims to enhance the existing tooled metodology called QuaRTOS-DSE by improving the formalization and the implementation of the existing methodology. The internship will address the exploration and the evaluation of complex critical SW architecture. Obtained SW architecture will be evaluated by a formal verification of extra functional system properties using existing tools. The exploration and the evaluation of complex critical SW architecture will be performed with an Iterative tool (a first version with a first formalization of the approach exists), at the level of functions, tasks, agents, actors and will integrate some SotA architecture strategies and best practices for critical SW. The approach must integrate an evaluation of some metrics and a connection with evaluation tools. The existing framework has very slight integration of the HW model, limitations on construction of input model (abstraction level) and limitations model transformation/generation for the...

Master's degree, Bac +5 - Master of Science Understanding embedded critical SW, and knowledge of formal methods would be a plus. English fluent, teamwork, curiosity In line with CEA's commitment to integrating people with disabilities, this job is open to all.
Durée (Mois):	6

15 nov., 2025

STAGE, Stage

Dynamic Distribution Shifts Ood Detection With Dynamic Thresholds H/F

CEA 48.7126483,2.1934657

The detection of out-of-distribution (OoD) samples is crucial for deploying deep learning (DL) models in real-world scenarios. OoD samples pose a challenge to DL models as they are not represented in the training data and can naturally arrive during deployment (i.e., a distribution shift), increasing the risk of obtaining wrong predictions. Consequently, OoD samples detection is crucial in safety-critical tasks, such as healthcare or automated vehicles, where trustworthy models are required. The existing literature for the OoD detection problem focuses on the development of confidence scores where a threshold is applied to build a binary classifier to tell if a sample is in-distribution (InD) or OoD. In particular, the confidence score threshold is typically set using the values that correspond to InD samples, such that 95% of the confidence score values from InD samples fall above the selected thresholds, i.e., 95% True Positive Rate. However, setting a fixed threshold can lead...

Master students (M1/M2 - France) Proficiency in Python, NumPy, SciPy, sciki-tlearn, PyTorch, Solid background in math, probability & statistics

15 nov., 2025

STAGE, Stage

Design Of a Reinforcement Learning-Driven Scheduler For Efficient And Frugal Container Orchestration H/F

CEA 48.7126483,2.1934657

Objective: The goal of this internship is to design and evaluate a new intelligent scheduling strategy using reinforcement learning (RL). The idea is to enable the system to learn how to make smarter scheduling decisions over time, optimizing container placement and sizing, dynamic resource allocation, response time and energy consumption and even inter-container dependencies such as shared data or communication patterns. Your missions: During this internship, you will: Explore and understand the orchestration framework developed within the team. Conduct a state-of-the-art study on RL-based scheduling in cloud and distributed environments. Design, implement, and train a new RL-based scheduler. Develop a feature extraction module to characterize container behavior and guide the RL agent's decisions. Evaluate your approach through experiments and benchmark comparisons

Profile sought We are looking for a motivated student in the final year of a Master's or Engineering program in Computer Science, Artificial Intelligence, or a related field, with: Good programming skills (Python preferred). Interest in machine learning and distributed systems. Curiosity, creativity, and strong problem-solving abilities.

14 nov., 2025

STAGE, Stage

Runtime Root-Cause Analysis For Intelligent Robots Via Causal ai Techniques H/F

CEA 48.7126483,2.1934657

Root-Cause Analysis (RCA) is a systematic process for identifying the fundamental cause of a problem or failure, rather than merely addressing its symptoms. It aims to understand why something went wrong in order to take appropriate actions and prevent recurrence. RCA is essential for robots that operate outside strictly controlled environments, where they are inevitably confronted with unexpected situations and failures. Symptoms can range widely, including erratic movements, sudden halts, or suboptimal task outcomes. RCA distinguishes these symptoms from the actual causes, which may include hardware or software bugs, inaccurate behavior specifications, or environmental factors. By pinpointing the root cause, robots can select appropriate goals for repair or system adjustments. This informed decision-making enhances resilience and ensures long-term safe autonomy for robots. Causal inference is a branch of AI research that focuses on understanding and modeling cause-and-effect...

The candidate should be undergoing a master (or equivalent) in computer science, robotics, embedded systems or closely related topics. The identified skills are: Strong programming skills in Python, with experience in data analysis and machine learning libraries. Familiarity with probabilistic modelling and Bayesian networks, including causal inference techniques is an advantage. Experience with Docker, CI/CD, and GitLab, as well as with robotics simulation environments and ROS 2, is an advantage. Self-learning and teamwork skills, motivation and interest to work in an interdisciplinary environment. Excellent communication skills in English, international candidates are encouraged to apply, knowledge of the French language is not required.

13 nov., 2025

STAGE, Stage

Learning To Focus Physics-Informed Deep Learning For Super-Resolved Ultrasonic Phased-Array Imaging H/F

CEA 48.7295843,2.1483258

Ultrasonic phased-array imaging is a core technology in non-destructive testing (NDT) for detecting defects such as cracks or voids in industrial components. By electronically steering ultrasonic beams, phased arrays generate detailed 3D images of internal structures. The Total Focusing Method (TFM) is the standard reconstruction algorithm, achieving diffraction-limited resolution by coherently summing signals from all emitter-receiver pairs. However, conventional TFM suffers from key limitations: its resolution is constrained by diffraction and array pitch, grating lobes degrade image quality, and it assumes uniform sound velocity. It also struggles to resolve sub-wavelength defects, limiting its effectiveness in complex or heterogeneous materials. Recent deep learning methods have improved ultrasonic imaging through denoising and super-resolution, but most operate as black boxes without physical interpretability. They often fail to generalize across array geometries or material...

The ideal candidate will have a Master's degree in Electrical Engineering, Applied Physics, Computer Science, or a related discipline. A strong background in signal and image processing, deep learning (PyTorch, TensorFlow), and programming in Python is expected. Prior experience with acoustic or ultrasonic imaging, inverse problems, or physics-informed machine learning will be considered a strong advantage.

05 nov., 2025

STAGE, Stage

S3 - Navigation And Manipulation With a Mobile Robot And Arm And Llm Interaction H/F

CEA Île-de-France

Missions: During this internship, you will explore integration of 3D reconstruction and natural language commands via large language models for a smart mobile robot (robot base + manipulator arm). The internship may notably include (depending on task advancement): -Reviewing existing literature on the subject -Downloading and testing the best approaches identified -Implementing a pipeline in simulation Implementing a pipeline on a real robot Job-related benefits Joining the CEA List and the LVML as an intern means: Working in one of the most innovative research organizations in the world, addressing societal challenges to build the world of tomorrow Discovering a rich ecosystem: privileged connections between the industrial and academic sectors Conducting research autonomously and creatively: encouragement to publish results (scientific articles, patents, open-source codes) Join a young and dynamic team Benefit from an internal computing infrastructure with more than 300...

Qualifications: Students in their 5th year of studies ( Master 2) or gap year Computer vision skills (3D vision, image processing) Python proficiency in a deep learning framework (preferably PyTorch

03 nov., 2025

STAGE, Stage

Backdoor Attack Scalability And Defense Evaluation In Large Language Models H/F

CEA 48.7295843,2.1483258

Context: Large Language Models (LLMs) deployed in safety-critical domains face significant threats from backdoor attacks. Recent empirical evidence contradicts previous assumptions about attack scalability: poisoning attacks remain effective regardless of model or dataset size, requiring as few as 250 poisoned documents to compromise models from up to 13B parameters. This suggests data poisoning becomes easier, not harder, as systems scale. Backdoors persist through post-training alignment techniques like Supervised Fine-Tuning and Reinforcement Learning from Human Feedback, compromising current defenses. However, persistence depends critically on poisoning timing and backdoor characteristics. Current verification methods are computationally prohibitive-Proof-of-Learning requires full model retraining and complete training transcript access. While step-wise verification shows promise for runtime detection, scalability to production models and resilience against adaptive adversaries...

Requirements: Background in computer science or a related field, with a focus on machine learning security, or adversarial machine learning. Strong programming skills in languages commonly used for machine learning tasks (e.g., Python, C++). Experience with machine learning systems, model training, or adversarial robustness is a plus. Ability to work independently and collaborate in a research-driven environment. Comfortable working in English, essential for documentation purposes.

31 oct., 2025

STAGE, Stage

Open-Set Object Detection Challenging Vlm To Understand Unknown Objects & Contexts H/F

CEA 48.7126483,2.1934657

Your missions within this internship are to: - Study state-of-the-art methods of Open Set Object Detection (OSOD) as well as Visual Language Models (VLM) in the context of Open World containing both known and unknown objects; - Design an object detector aware of the existence of the unknown, and able to describe the unknown by comparing it to or distinguishing it from what is known via certain characteristics that can be described textually; - Evaluate the proposed method on recent OSOD benchmarks and compare to the state of the art; - Challenge these methods by applying them to new contexts (e.g. aerial images, medical imaging); - If relevant, submit your contributions to an international conference or workshop for publication. Join CEA List and LVA as an intern to: - Work in one of the most innovative research organizations in the world, addressing societal challenges to build the world of tomorrow - Discover a rich ecosystem: privileged connections between the industrial and...

Students in their 5th year of studies (M2) - Computer vision skills - Machine learning skills (deep learning, VLM) - Python proficiency in a deep learning framework (especially PyTorch)

24 oct., 2025

STAGE, Stage

Hyperspectral Imager Implementation And Testing H/F

CEA 48.7126483,2.1934657

Please view a better description on: https://kalisteo.cea.fr/index.php/offres-demploi/ CEA recently invented a new 3D hyperspectral imager for moving objects. This imager requires specific reconstruction code to operate. We have the theoretical formulation of the solution but still lack a proper implementation and testing. During this internship, you will develop code to exploit this imager for 3D reconstruction and spectral information extraction. You may have to propose improvements to the optical system. Reviewing existing literature on the subject Implementing the base 3D+spectral reconstruction code Producing scientific content (publications, patents, databases, etc.)

24 oct., 2025

STAGE, Stage

Object Model-Conditioned Image Detection And Segmentation H/F

CEA 48.7126483,2.1934657

See illustrations on: https://kalisteo.cea.fr/index.php/offres-demploi/ During this internship, you will investigate state of the art techniques for detection of objects in images when the CAD model of the object is available (CAD-conditioned detection). Indeed, most state of the art approaches start with an object-agnostic segmentation (using Meta's SAM for instance) followed by template matching. But this does not work in industrial contexts, where objects are not cars, dogs or other everyday classes easily distinguishable from the background. The internship will notably include: Reviewing existing literature on the subject Downloading and testing the best approaches identified Implementing existing approaches lacking proper open-source implementation Creating, implementing and benchmarking new approaches

22 oct., 2025

STAGE, Stage

Design Of Fault Injection Models Within Pre-Silicon Security Methodologies H/F

CEA Île-de-France

µArchiFI [2,3] is one of these pre-silicon tools, it constructs a formal transition system from a Verilog processor description, a binary program, and an attacker model that encodes the fault model. However, the fault models used by µArchiFI do not incorporate layout information. Analyses are performed at the Register Transfer Level (RTL) and can evaluate a wide range of fault models (bit/word set, reset, flip, and symbolic behaviors) on signals selected individually. In a real fault attack scenario, for instance, using a laser source as the fault injection tool, it may hit different bits of the same signal or of different signals. Goals The internship objective is to enhance µArchiFI with new fault models so that signals that are affected by the laser beam are selected according to laser-spot location regarding the circuit layout. This requires: 1) integrating layout information and location constraints into the fault models, 2) modelling the laser beam's Gaussian profile to...

Profile. This position is aimed at students seeking an ambitious technical internship, eager to gain significant experience in industry-related technological research. It is particularly well-suited to students considering a doctorate, with new funded positions offered each year within the department. The internship is aimed at students in their final year of engineering school (or Master 2) in computer science or microelectronics, or equivalent levels, preferably with a specialization in processor systems/architecture or formal methods. Knowledge of micro-architecture or cybersecurity is an asset, but not a prerequisite. A strong capacity for personal work, ability to work in a team and motivation to take on technical challenges are essential. Programming capabilities, in particular in C++ and Object-Oriented Programming (OPP) Opportunities. Practical application: work on state-of-the-art pre-silicon tools to assess the security of secure processors against fault-injection...

15 oct., 2025

STAGE, Stage

Camera-Radar 3D Perception Model For Autonomous Driving H/F

CEA Île-de-France

The goal of this internship is to investigate advanced methods for fusing complementary information from cameras and radars to achieve a robust and accurate 3D understanding of the driving environment (3d detection, 3d semantic segmentation). Cameras provide rich semantic and structural cues for object recognition and scene interpretation, while radars deliver precise range and velocity measurements that remain reliable under adverse weather and lighting conditions. By combining these modalities, the intern will explore novel deep learning architectures for multi-sensor fusion, with the aim of improving accuracy in complex driving scenarios. To achieve these objectives, the intern will be expected to: - Review the state of the art on camera-radar fusion for 3D perception - Design, develop and evaluate a novel deep learning models for multi-sensor fusion - Contribute to research reports and potential publications...

15 oct., 2025

STAGE, Stage

Développement et Intégration d'Une Tsn Application Function Af en 5G H/F

CEA 48.7126483,2.1934657

Version française Pour répondre aux exigences de l'Industrie 4.0 (robotique, automatisation), l'intégration du standard Time-Sensitive Networking (TSN) aux réseaux 5G est devenue une priorité. Le système 5G doit garantir un service déterministe en agissant comme un pont TSN. Ce stage se concentre sur le développement d'un composant logiciel clé : la TSN Application Function (TSN-AF), chargée de traduire les exigences de communication industrielle en politiques de Qualité de Service (QoS) pour le coeur de réseau 5G (5GC). Les travaux seront réalisés sur la plateforme open-source de référence OpenAirInterface (OAI). L'objectif principal est de concevoir, développer et valider la fonction TSN-AF et de l'intégrer au coeur de réseau 5G OAI. Les tâches clés sont : - Analyse de Spécifications : étude approfondie des standards 3GPP concernant l'intégration 5G-TSN pour maîtriser les mécanismes de QoS, les modèles d'architecture et les interfaces requises.. - Conception de l'Architecture :...

Vous êtes en formation pour un diplôme d'ingénieur ou master 2 en réseau informatique. Vous appréciez travailler en équipe mais savez être autonome dans vos missions. Vous êtes ouvert aux nouvelles expériences et vous êtes force de proposition. La connaissance ou une expérience des outils suivant est un plus : Langages C/C++, Python Systèmes Linux Environnement conteneurisé (Docker/Kubernetes) Bonnes connaissances sur l'architecture 5G, SDN, le protocole Netconf Conformément aux engagements pris par le CEA en faveur de l'intégration de personnes en situation de handicap, cet emploi est ouvert à tous et toutes. In line with CEA's commitment to integrating people with disabilities, this job is open to all.

13 oct., 2025

STAGE, Stage

Image Editing Of Complex Visual Scene Via Natural Language H/F

CEA Île-de-France

This internship focuses on the emerging field of natural language-guided image editing, specifically targeting the generation and modification of complex scenes based on verbal descriptions. The candidate will work on designing and implementing novel methods that can interpret natural language to manipulate or generate detailed images representing multifaceted scenarios (e.g., crowd scenes, cityscapes, interactions between multiple objects). This project presents several key challenges, including: -Scene Complexity: Managing multiple objects and their relationships in a scene adds significant complexity. The goal is to maintain coherence and accuracy in the edited images, even when the scenes described involve intricate interactions between various elements. -Multimodal Integration: Successfully combining linguistic and visual inputs to obtain visual outputs, is a complex problem requiring seamless interaction between natural language processing (NLP) and computer vision models....

-Students in their 5th year of studies (M2) -Computer vision skills -Machine learning skills (deep learning, LLM, VLM, generative AI) -Python proficiency in a deep learning framework (especially PyTorch or TensorFlow)

20 offres internship trouvées