CV
Curriculum updated on 06/05/2025
Basics
Name | Ariel Gjaci |
Label | Postdoctoral Researcher |
ariel.gjaci@iit.it | |
Phone | +393483776041 |
Url | https://arielgjaci.com |
Summary | Postdoctoral Researcher with a multidisciplinary background in biomedical engineering, robotics, and artificial intelligence. My work spans computer vision, machine learning, multimodal systems, robotics, and computational acoustics, with applications in social robotics and immersive environments. I am driven by the goal of advancing AI research across diverse domains. |
Work
-
2025.02 - now Postdoctoral Researcher
Italian Institute of Technology (IIT), Genova, Italy
I am working with the Pavis Team on the European project XTREME, which focuses on mixed reality solutions for art and culture, integrating multimodal audio and vision processing . During this period, I will collaborate with a team to develop a novel audio-visual-visual model capable of rendering realistic binaural audio at any position within a room. This model will then be integrated into a mixed reality system to deliver immersive audio and video experiences.
- Writing a survey paper on learning-based methods for novel-view audio rendering
- Developing a new method for high-quality binaural audio rendering using visual features.
- Collaborating with researchers across Europe.
- Collecting real-world multimodal data (e.g., museums, orchestras) for training audio rendering models.
- Contributing to the release of a production-level system.
-
2021.05 - 2021.11 Software Engineer Intern
Akka Technologies, Milano, Italy
Worked in a team project for developing the back-end software for managing aircraft panels.
- Developed a software system for managing military aircraft panels.
- Delivered and presented a working demo to stakeholders.
-
2021.01 - 2025.05 PhD Candidate
University of Genoa, Genova, Italy
The goal of my PhD was to create a hybrid rule-based and data-driven approach to generate culture-aware co-speech gestures on social robots. The rule-based part defines heuristic rules based on semantic scores from Cross-Encoder language models. The data-driven part generates co-speech gestures by implementing a multimodal, culture-aware generative model based on Diffusion-Transformers. The cultural component is analyzed from multimodal data and incorporated into the generated gestures, ensuring that the non-verbal communication of social robots is contextually and culturally appropriate.
- Developed a large-scale multimodal dataset.
- Applied ML models (SVMs, Transformers, MLPs, VQ-VAEs, LLMs) to analyze cultural factors.
- Developed subject-independent cultural embeddings using Fishr and Adversarial Learning.
- Built rule-based algorithms for identifying gestures from text using Semantic Similarity.
- Built a Diffusion-Transformer model for multimodal gesture generation.
- Reviewed papers for AI and Robotics journals and conferences.
- Co-organized a workshop at ECCV.
- Collaborated with researchers of King's College London.
- Obtained certifications from Coursera and international summer schools.
- Chaired a session of IEEE RO-MAN conference.
Education
-
2022.01 - 2023.04 London
-
2021.01 - 2025.05 Genoa
-
2018.01 - 2021.06 Genoa
Master's Degree in Robotics Engineering
University of Genoa
Robotics
- Computer vision
- Machine learning
- Multivariable and Non-Linear Control Theory
- Embedded Systems
- ROS
- Real-Time Operating Systems
- Mobile Robotics
- Manipulators
- Mechanical Design
- Social Robotics
- Biomedical Robotics
-
2014.01 - 2018.07 Genoa
Bachelor's Degree in Biomedical Engineering
University of Genoa
Biomedical Engineering
- Mathematics
- Geometry
- Physics (Thermodynamics, Mechanics, Electromagnetism
- Analog and Digital Signal Processing
- Electromagnetic Fields
- Biomedical Instrumentation
- Chemistry
- Materials Science
- Biomechanics
- Physiology
- Programming (C / C++ / Matlab)
- Electronic Circuits
- Bioelectronics
Certificates
1st Doctoral Summer School on Robotics and Intelligent Machines | ||
Scuola Superiore Sant'Anna | 2023-10-01 |
Topics in Modern Machine Learning | ||
MaLGa - Machine Learning Genoa Center | 2023-06-01 |
Natural Language Processing with Classification and Vector Spaces | ||
Coursera | 2022-04-01 |
Skills
Programming | |
Python | |
C++ | |
C | |
R | |
Matlab | |
Kotlin |
ML Frameworks | |
Pytorch | |
Tensorflow | |
Scikit-learn | |
Pandas |
Operating Systems | |
Linux | |
Windows | |
ROS |
Embedded Systems | |
dsPICDEM 2 | |
Raspberry pi | |
Arduino |
Soft Skills | |
Motivation | |
Learning | |
Teamwork | |
Leadership | |
Problem Solving | |
Communication | |
Time Management |
Languages
Italian | |
Native speaker |
Albanian | |
Native speaker |
English | |
Fluent |
Interests
Machine Learning | |
Generative models | |
Multimodal models | |
Large Language models | |
Domain Generalization | |
Quantum Communication | |
Reinforcement Learning |
Robotics | |
Social Robots | |
Multimodal Interactions |
Computer Vision | |
Motion Generation | |
Vision Language Models | |
Object Identification |
Computational Acoustics | |
Acoustic Fields | |
Binaural Audio | |
Audio Rendering |
Medicine | |
ML-based diagnosis | |
Drug Discovery |