cv

Research Engineer in AI

CEA-LIST, DILS, LSEA

Research in Deep Learning Uncertainty, Out-of-Distribution Detection, and Trustworthy AI/AI Safety.

• Trustwrothy Deep Learning Team Leader
• Participation in the EU&FR funded projects: Comp4Drones (WP4-T4.4 leader and contributor), Confiance.ai program (contributor), DeepGreen (contributor)

Data Scientist

Bamboo Tec

Enterprise data analysis and visualization (dashbords)

• Data Science & Engineering Team Leader

Research Assistant

University of Applied Arts and Science of Souther Switzerland - SUPSI

Embedded Systems and IoT R&D

• Embedded Machine Learning
• IoT protocol design and implementation
• RF PCB Design

UAV Research Engineer

Jalasoft

UAV Computer Vision, Navigation, and Embedded Systems development

• NIR image processing for NDVI
• power transmission lines detection
• ROS integration and PX4 autopilot development

Adjunt Lecturer and Researcher

Universidad Privada Boliviana

Research on Smart Embedded Systems for IoT, and Lecturer from the Electronics and Computer Science majors

• Smart Street Lightning Project
• Courses Taught: Embedded Electronic Systems (2016), Microprocessor Architecture and Technology (2015, 2016, 2019), Electronics and Telecom. Project (2018), Telecom. Electronics (2018), Electronic Instrumentation (2014, 2015)

Ph.D. in Computer Science - Artificial Intelligence

Universite Paris-Saclay

Deep Learning Uncertainty Quantification, Trustworthy AI, AI-Safety

• Ph.D. Thesis Title: Deep Neural Network Uncertainty Runtime Monitoring for Robust and Safe AI-based Automated Navigation

Master of Science in Engineering - Embedded Sytems and Microelectronics

University of Applied Arts and Science of Souther Switzerland - SUPSI

Precision Systems and Telecom

• MSc. Thesis Title: Real-Time Human Footstep Recognition on Smart Anti-Static Floor

B.S in Electronics and Telecommunications Engineering

Universidad Privada Boliviana

Embedded Systems and IoT

• Grade Project Title: VIRMS – A Vehicle Information and Road Monitoring System

RETECA Fondation 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.

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

ArXiV

This paper shows that the current evaluation protocol for OOD-OD violates the assumption of non-overlapping objects with respect to the In-Distribution (ID) datasets, and obscures crucial situations such as ignoring unknown objects, potentially leading to overconfidence in deployment scenarios where truly novel objects might be encountered. To address these limitations, we manually curate, and enrich the existing benchmark by exploiting semantic similarity to create new evaluation splits categorized as near, far, and farther from ID distributions. Additionally, we incorporate established metrics from the Open Set community, providing deeper insights into how effectively methods detect unknowns, when they ignore them, and when they mistakenly classify OOD objects as ID. Our comprehensive evaluation demonstrates that semantically and visually close OOD objects are easier to localize than far ones, but are also more easily confounded with ID objects. Far and farther objects are harder to localize but less prone to be taken for an ID object.

Latent Representation Entropy Density for Distribution Shift Detection

PMLR

This paper proposes using simple sample-based techniques for estimating uncertainty and employing the entropy density from intermediate representations to detect distribution shifts. We demonstrate the effectiveness of our method using standard benchmark datasets for out-of-distribution detection and across different common perception tasks with convolutional neural network architectures. Our scope extends beyond classification, encompassing image-level distribution shift detection for object detection and semantic segmentation tasks.

Towards Dependable Autonomous Systems Based on Bayesian Deep Learning Components

IEEE

This paper provides a method that considers the uncertainty and the interaction between BDL components to capture the overall system uncertainty. We study the effect of uncertainty propagation in a BDL-based system for autonomous aerial navigation. Experiments show that our approach allows us to capture useful uncertainty estimates while slightly improving the system’s performance in its final task. In addition, we discuss the benefits, challenges, and implications of adopting BDL to build dependable autonomous systems

Quantifying and Using System Uncertainty in UAV navigation

ArXiV

We study the effect of the uncertainty from perception representations in downstream control predictions. Moreover, we leverage the uncertainty in the system's output to improve control decisions that positively impact the UAV's performance on its task.