I specialize in crafting practical and effective deep models tailored for real-world usability, with a primary focus on solving intricate challenges related to motion problems
Deep/Machine Learning Expertise:
Proven track record in developing real-time deep learning models for tasks such as sensor data fusion and motion/trajectory analysis using various sensors.
Emphasis on creating solutions with real-time functionality, low memory footprint on edge devices, and adherence to power requirements.
Applied experience in different domains such as AR/VR/XR, self-driving vehicles and robotics.
Understanding of edge devices, embedded systems and HW/SW integration.
Heavy expertise in developing 0->1 POC (Proof of concept).
Spatio-temporal and Motion Analysis:
Specialization in real-time deep models for multiple-agent motion, action, and trajectory forecasting/understanding, considering uncertainty in predictions.
Research focuses on devising novel metrics for evaluating the performance of such models.
Understanding of control theory and motion dynamics.
Outdoor and Indoor Localization:
Industrial experience in non-visual localization employing classical and data-driven approaches.
Pioneered the development of multi-level data fusion systems, ensuring robust localization solutions.
Proficient with IMUs, magnetometers, and ultrasonic technologies for developing heading and tracking systems.
Integrative approach using both physics and data-driven solutions.
Familiarity with GNSS and similar positioning systems.
Author of 600+ citations, with publications in renowned venues such as CVPR, ECCV, and AC.
Active involvement as a reviewer for top-tier conferences.
Co-supervision of Ph.D. students engaged in multidisciplinary research.
Skilled in Python, deep learning libraries, C++, Matlab, and software engineering.