cv

Research Engineer in AI — Trustworthy/Safe Deep Learning Team Leader

CEA-List, Université Paris-Saclay, Laboratory of Embedded and Autonomous Systems (LSEA)

Research on DNNs/LLMs/VLMs/VLAs confidence representation (uncertainty estimation) for out-of-distribution/hallucination detection, runtime monitoring in complex computer vision tasks (object detection demo), quantization and embedded uncertainty estimation, safe and robust end-to-end DNN-based UAV navigation, world models, and AI Safety.

• Ph.D. thesis supervision: Deep Neural Network Uncertainty Estimation on Embedded Targets (Aymen Bouguerra, 2024 - present); Out-of-Distribution Detection with Vision Foundation Models and Post-hoc Methods (starting Oct. 2026); internships supervision
• Developed RunIA-Core, a Python library for DNN confidence representation and runtime monitoring, with support for PyTorch and PyTorch-Lightning, and integration with MLFlow for experiment tracking
• Contributed to the PRIMaL project: developed an uncertainty-based method for LLM extrinsic & intrinsic hallucination detection, developed OoD object detection methods, and collected a dataset for UAV navigation using the PyBullet simulator (gym-pybullet-drones)
• Contributing to the DeepGreen project for embedded AI (France 2030 program): developed a dropout operator for the Aidge library for embedded DNN uncertainty estimation, and studied the reliability & robustness of quantized CLIP models
• Contributed to the Confiance.ai program (France 2030 program): developed an OoD detection method library
• Contributed to the Comp4Drones project (H2020 program, European Union): led WP4-Task 4.4 on UAV runtime monitoring, and developed a distributed uncertainty runtime monitoring system for DNN-based UAV navigation systems

Data Scientist

Bamboo

Enterprise data analysis & data visualization

• Engineering team leader
• Implemented PowerBI dashboards to support enterprise decision-making

Research Assistant

University of Applied Sciences and Arts of Southern Switzerland (SUPSI), ISEA, SMT Lab

Part-time embedded systems development

• Developed embedded machine learning prototypes for signal classification
• Developed embedded system prototypes for wireless IoT applications (LoRaWAN, 802.15.4, Bluetooth)
• RF PCB design

Part-time Researcher and Adjunct Lecturer

Universidad Privada Boliviana (UPB)

Research in embedded wireless IoT applications; lecturer

• Contributed to the Smart Street Lighting project (SRESLi)
• Courses taught: Embedded Electronic Systems (2016), Microprocessor Architecture & Technology (2015, 2016, 2019), Electronics & Telecom. Project (2018), Telecom. Electronics (2018), Electronic Instrumentation (2014, 2015)

Unmanned Aerial Systems Researcher

Jalasoft

Localization & navigation, ROS integration, embedded systems integration, computer vision tasks

• Built a custom UAV from scratch employing the PX4/Pixhawk autopilot
• Integrated software & hardware components (modular system) using ROS, and implemented onboard computer vision tasks

Ph.D. in Computer Science - Artificial Intelligence

Université Paris-Saclay

Deep Learning Uncertainty Quantification, Trustworthy AI, AI-Safety

• Thesis: Deep Neural Network Uncertainty Runtime Monitoring for Robust and Safe AI-based Automated Navigation
• DNN uncertainty-based out-of-distribution detection, uncertainty propagation for robust UAV navigation, confidence monitoring in UAV learning-based components for trustworthy navigation

MSc. in Engineering: Embedded Systems & Microelectronics

University of Applied Sciences and Arts of Southern Switzerland (SUPSI)

Precision Systems and Telecom

• Thesis: Real-Time Human Footstep Recognition on Smart Anti-Static Floor
• Signal processing and classification using machine learning targeting resource-constrained hardware

BSc. in Electronics and Telecommunications Engineering

Universidad Privada Boliviana (UPB)

Embedded Systems and IoT

• Grade project: VIRMS: A Vehicle Information and Road Monitoring System (paper)
• Embedded systems, RTOS, data acquisition & processing, IoT, embedded GUI development

Disentangling the Good From the Bad: Quantization-Induced Flips Are Not Random

Conference on Neural Information Processing Systems (NeurIPS), to appear

Bouguerra, A., Gomez-Villa, A., Mraidha, C., Arnez, F.

Drive-SynOOD-OD: A Diffusion-Inpainted Synthetic Out-of-Distribution Object Detection Benchmark for Autonomous Driving

IEEE International Conference on Vehicular Electronics and Safety (ICVES), to appear

Montoya, D., Espinoza Mayzer, M. S., Arnez, F.

Quantization and Corruption Robustness in Deployed Road-Scene Perception

IEEE International Conference on Vehicular Electronics and Safety (ICVES), to appear

Bouguerra, A., Radermacher, A., Mraidha, C., Arnez, F.

Less Precise Can Be More Reliable: A Systematic Evaluation of Quantization's Impact on VLMs Beyond Accuracy

International Conference on Machine Learning (ICML)

Bouguerra, A., Montoya, D., Gomez-Villa, A., Mraidha, C., Arnez, F.

The SIGReg Objective as Variational Free Energy: A Theoretical Active-Inference Account of JEPA World Models

arXiv preprint arXiv:2607.13612

Arnez, F., & Gomez-Villa, A.

Digital Twins and World Models: A Systematic Taxonomic Disambiguation

International Conference on Advanced Information Systems Engineering (CAiSE) Workshops, pp. 61-66

Suri, K., & Arnez, F.

The Map of Misbelief: Tracing Intrinsic and Extrinsic Hallucinations Through Attention Patterns

Proceedings of the AAAI Symposium Series (AAAI Fall Symposium - ATRACC), 7(1)

Hajji, E., Bouguerra, A., Arnez, F.

FindMeIfYouCan: Bringing Open Set Metrics to Near, Far and Farther Out-of-Distribution Object Detection

arXiv preprint arXiv:2506.14008 (under review)

Montoya, D., Bouguerra, A., Gomez-Villa, A., & Arnez, F.

Latent Representation Entropy Density for Distribution Shift Detection

Conference on Uncertainty in Artificial Intelligence (UAI)

Arnez, F., Vasquez, D. A. M., Radermacher, A., & Terrier, F.

Towards Dependable Autonomous Systems Based on Bayesian Deep Learning Components

2022 18th European Dependable Computing Conference (EDCC), pp. 65-72, IEEE

Arnez, F., Espinoza, H., Radermacher, A., & Terrier, F.

Improving Robustness of Deep Neural Networks for Aerial Navigation by Incorporating Input Uncertainty

Workshop on Artificial Intelligence Safety Engineering (WAISE 2021), International Conference on Computer Safety, Reliability, and Security (SAFECOMP), pp. 219-225, Springer

Arnez, F., Espinoza, H., Radermacher, A., & Terrier, F.

A Comparison of Uncertainty Estimation Approaches in Deep Learning Components for Autonomous Vehicle Applications

Proceedings of the Workshop on Artificial Intelligence Safety 2020 (AISafety), CEUR-WS vol. 2640

Arnez, F., Espinoza, H., Radermacher, A., & Terrier, F.

Full list of publications in Google Scholar profile

Google Scholar

Workshop & Conference Organizing Committee Member

Conferences and workshops

Workshop on AI Safety and Security at IJCAI 2024; IEEE International Conference on Vehicular Electronics and Safety (ICVES) 2026 (ongoing)

Reviewer

Conferences and workshops

NeurIPS (2026), ICML (2026, Gold Reviewer), ICLR (2027, ongoing), BMVC (2026), CVPR (2025), AAAI (2026, 2025), UAI (2026, 2025, 2024), WACV (2025), IV (2025, 2022), AI Safety workshop at IJCAI (2024), Releasing Robots into the Wild workshop at ICRA (2022)

Gold Reviewer, ICML 2026

International Conference on Machine Learning (ICML)

Recognized among the top reviewers of ICML 2026.

RETECA Foundation Scholarship

RETECA Foundation, Switzerland

The aim of the RETECA Foundation is to help and support young researchers from developing countries engaged in post-education or research activities in the field of electrical engineering or applied electrical engineering.