CV

Education

• Ph.D in Engineering, Faculty of Engineering , University of Buenos Aires, 2020
  • Thesis: Early Language Acquisition in Cortical Dynamics, a Computational Approach.
  • Biologically inspired modelling for early language acquisition phenomena explanation.
  • High Performance Computing (HPC) for science. Project running on Cooley resource at Argonne Leadership Computing Facility ( ALCF ).
• B.S. in Mendoza Regional Faculty , National Technological University, 2012
  • Senior Thesis: Kalman Filter Implementation in FPGA, Reconfigurable Computation Laboratory.

Work experience

• May 2025 - present: AI Software Developer, Active Inference Team
  • StanhopeAI, London, United Kingdom
  • Duties include:
    • Developing probabilistic autonomy software for drone systems operating in GNSS-denied environments.
    • Applying Active Inference and the Free Energy Principle to perception, belief updating, planning, and action selection under uncertainty.
    • Designing state-space representations and generative models that integrate sensory evidence with predictive dynamics.
    • Implementing and evaluating autonomy algorithms in Python, MATLAB, and C/C++.
    • Contributing to multi-agent and swarm autonomy concepts for information sharing, coordinated decision making, and mission behavior under incomplete observations.
• Nov 2021 - Apr 2025: Postdoctoral researcher
  • Northwestern‑Argonne Institute of Science and Engineering ( NAISE )
  • Duties included:
    • Self-supervised learning through Vision Transformers applied to label-scarce edge computing environments.
    • Self-supervised learning for full-sky cloud image characterization and ecosystem activity monitoring.
    • Federated and distributed learning workflows across edge devices and leadership-class HPC resources.
    • Development of PTZJEPA, a closed-loop active perception framework that combines a predictive world model with reinforcement-learning control of a pan-tilt-zoom (PTZ) camera.
    • Deployment of multimodal vision-language models, including Florence-2 and LLaVA, on Dell XR2 servers with NVIDIA T4 GPUs for edge inference.
  • Supervisor: Nicola J. Ferrier
• Apr 2020 ‑ Nov 2021: Postdoctoral researcher
  • CONICET Mendoza Technological Scientific Center
  • Duties included:
    • Use of Deep Learning Frameworks for the implementation of self‑supervised contrastive learning of visual features through active foveated saccades.
    • Biologically inspired active foveated saccade visual system utilizing Transformers, ResNet and Deep Reinforcement Learning architectures.
    • Application of CNN and ResNet architectures for Nanopore Translocation feature extraction from noisy signals.
    • Application and adaptation of End‑to‑end object detection with Transformers (DETR) for detection of events in noisy signals
  • Supervisor: Alejandro Wainselboim
• Apr 2012 ‑ Apr 2013: Research Intern
  • National Technological University
  • Duties included: Kalman Filter Numerical Tests in Matlab. Implementation in VHDL of resource and delay efficient matrix multiplication algorithms.
  • Technical Skills: Very High Speed Integrated Circuit (VHSIC) Hardware Description Language (VHDL), Matlab.
  • Supervisor: Rodrigo Gonzalez

Skills

• Autonomy and Robotics
  • Autonomous perception, GNSS-denied navigation, active vision
  • Predictive world models, uncertainty-aware decision making, embodied intelligence
  • Active Inference, Free Energy Principle, reinforcement learning
• Programming
  • Python (PyTorch, NumPy, Scikit‑learn, Pandas, DistributedDataParallel, h5py, etc.)
  • C/C++, MATLAB, Nvidia DALI
  • OpenMP, MPI, HDF5
• Deep Learning and Vision
  • Vision Transformers, CNNs, ResNet, DETR, DINO, VICReg, I-JEPA/JEPA
  • Florence-2, LLaVA, object detection, multimodal learning
• Distributed, Edge, and HPC Systems
  • PyTorch DistributedDataParallel, mpi4py, multi-GPU training
  • ALCF Polaris, ThetaGPU, Cooley
  • Edge inference on NVIDIA GPU platforms, Dell XR2/T4, CPU fallback deployment
• Miscellaneous
  • Linux, Shell (Bash)
  • LATEX(Overleaf/Markdown)
  • Git, GitHub, GitLab, CMake
  • Docker, Singularity/Apptainer, Venv
• Soft Skills
  • Time Management, Teamwork, Problem‑solving
  • Documentation, Engaging Presentation
  • Student advising

Publications

A semi-supervised learning method for the identification of bad exposures in large imaging surveys

Yufeng Luo, Adam D. Myers, Alex Drlica-Wagner, Dario Dematties, Salma Borchani, Frank Valdes, Arjun Dey, David Schlegel, Rongpu Zhou, and DESI Legacy Imaging Surveys Team. (2026). "A semi-supervised learning method for the identification of bad exposures in large imaging surveys." Open Journal of Astrophysics. 9.

Perspective on self-supervised transformers for scalable data analysis, retrievals, and discovery in geosciences

Bhupendra A. Raut, Siddhant Gupta, Robert C. Jackson, Dié Wang, Adam Theisen, Jingjing Tian, Seongha Park, Joseph R. O’Brien, Maxwell A. Grover, Troy Arcomano, Nicki L. Hickmon, Scott M. Collis, Nicola J. Ferrier, and Dario J. Dematties. (2026). "Perspective on self-supervised transformers for scalable data analysis, retrievals, and discovery in geosciences." Artificial Intelligence for the Earth Systems. e250069.

Acoustic fingerprints in nature: A self-supervised learning approach for ecosystem activity monitoring

Dario Dematties, et al. (2024). "Acoustic fingerprints in nature: A self-supervised learning approach for ecosystem activity monitoring." Ecological Informatics. 83.

Containerization on a self-supervised active foveated approach to computer vision

Dario Dematties, Silvio Rizzi, and George Thiruvathukal. (2024). "Containerization on a self-supervised active foveated approach to computer vision." Revista Colombiana de Computación. 25, 29-38.

Doppler spectra from ARM Doppler Lidar during ARMing the Edge

Bhupendra Raut, et al. (2023). "Doppler spectra from ARM Doppler Lidar during ARMing the Edge." Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States). Atmospheric Radiation Measurement (ARM) Archive; Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States). Atmospheric Radiation Measurement (ARM) Data Center.

Adversarial Predictions of Data Distributions Across Federated Internet-of-Things Devices

Samir Rajani, et al. (2023). "Adversarial Predictions of Data Distributions Across Federated Internet-of-Things Devices." IEEE WF-IoT.

ARMing the Edge: Designing Edge Computing-capable Machine Learning Algorithms to Target ARM Doppler Lidar Processing

Robert C. Jackson, et al. (2023). "ARMing the Edge: Designing Edge Computing–Capable Machine Learning Algorithms to Target ARM Doppler Lidar Processing." Artificial Intelligence for the Earth Systems.

Let’s Unleash the Network Judgment: A Self-Supervised Approach for Cloud Image Analysis

Dario Dematties, et al. (2023). "Let’s Unleash the Network Judgment: A Self-Supervised Approach for Cloud Image Analysis." Artificial Intelligence for the Earth Systems. 2(2).

Optimizing cloud motion estimation on the edge with phase correlation and optical flow

Bhupendra A Raut, et al. (2023). "Optimizing cloud motion estimation on the edge with phase correlation and optical flow." Atmospheric Measurement Techniques. 16(5).

A Self-supervised Approach for Cloud Image Analysis

Bhupendra A Raut, et al. (2023). "A Self-supervised Approach for Cloud Image Analysis." 103rd AMS Annual Meeting.

Towards an Active Foveated Approach to Computer Vision

Dario Dematties, et al. (2022). "Towards an Active Foveated Approach to Computer Vision." Computación y Sistemas. 26(4).

A generalized transformer-based pulse detection algorithm

Dario Dematties, et al. (2022). "A generalized transformer-based pulse detection algorithm." ACS sensors. 7(9).

Neural correlates of semantic integration in Spanish verb-noun compounds: an ERP study

Mercedes Güemes, et al. (2022). "Neural correlates of semantic integration in Spanish verb-noun compounds: an ERP study." Language, Cognition and Neuroscience. 37(3).

A guide to signal processing algorithms for nanopore sensors

Chenyu Wen, et al. (2021). "A guide to signal processing algorithms for nanopore sensors." ACS sensors. 6(10).

Deep Learning of Nanopore Sensing Signals Using a Bi-Path Network

Dario Dematties, et al. (2021). "Deep Learning of Nanopore Sensing Signals Using a Bi-Path Network." ACS Nano. 15(9).

A Computational Theory for the Emergence of Grammatical Categories in Cortical Dynamics

Dario Dematties, et al. (2020). "A Computational Theory for the Emergence of Grammatical Categories in Cortical Dynamics." Frontiers in Neural Circuits. 14(12).

Neural Network Approach for Dielectric Characterization of Tissues in Microwave Frequencies using Coplanar Waveguide Transmission

Viktor Mattsson, et al. (2020). "Neural Network Approach for Dielectric Characterization of Tissues in Microwave Frequencies using Coplanar Waveguide Transmission." 2020 14th European Conference on Antennas and Propagation (EuCAP).

Towards High-End Scalability on Biologically-Inspired Computational Models

Dario Dematties, et al. (2020). "Towards High-End Scalability on Biologically-Inspired Computational Models." Advances in Parallel Computing. 36.

Phonetic acquisition in cortical dynamics, a computational approach

Dario Dematties, et al. (2019). "Phonetic acquisition in cortical dynamics, a computational approach." PLoS ONE. 14(6).

Talks

Towards Self-Supervised Learning @ the Edge

February 20, 2024

Talk at CS Seminar Series, Chicago, Illinois

Self‑Supervised Learning at the Edge

November 30, 2022

Talk at American Computing Machinery Chicago, Chicago, Illinois

Self‑Supervised Learning for Automated Species Detection from Passively Recorded Soundscapes in Avian Diversity Monitoring

November 15, 2022

Talk at Super Computing 2022, Dallas, Texas

Selected projects

• Predictive active vision: designed PTZJEPA, integrating joint-embedding predictive learning with autonomous pan-tilt-zoom control.
• GNSS-denied drone autonomy: developing Active Inference models for perception and decision making in aerial systems operating with uncertain or unavailable satellite positioning.
• Self-supervised vision at scale: implemented DINO, VICReg, and I-JEPA pipelines using Vision Transformers, distributed multi-GPU training, and environmental image datasets collected at edge sensing sites.
• Edge multimodal AI: deployed Florence-2 and LLaVA-based vision-language inference pipelines on constrained edge GPU systems for semantic interpretation of RGB and infrared imagery.
• Active foveated perception: built a reinforcement-learning visual agent that acquired self-supervised representations through sequential foveated observations.

Teaching

Linguistics and Experimental Neurobiology of Language

Linguistics and Experimental Neurobiology of Language

Mathematical Analysis II

Signals and Systems Analysis

Mathematical Analysis II

Service and leadership

• Nov 2023 - Feb 2025: Moderator and scientific coordinator for the Trillion Parameter Consortium (TPC) seminar series. TPC is an international community of nearly 1,000 participants from over 100 organizations including universities, research laboratories and institutes, and industry. TPC fosters collaborations toward the creation and responsible use of generative artificial intelligence in an open community and aspires to create a global network of resources and expertise that can help to grow that community.