Hey, I Am
Federico Fontana
A highly motivated and results-driven Computer Vision Ph.D. student with a strong focus on efficient deep learning and deepfake detection. My research explores architectural modifications and novel pruning and quantization techniques for transformers across various tasks. I am particularly interested in optimizing deep learning models through pruning, quantization, and distillation to enhance efficiency without sacrificing performance. My work extends to designing more compact and computationally efficient models for real-world applications.
With extensive experience in deep learning and AI, I have a strong engineering background, proficient in Python, C++, and CUDA. I specialize in developing and optimizing deep learning models, rapidly prototyping innovative solutions, and implementing cutting-edge techniques. I am eager to contribute my expertise to advancing efficient AI research and applications.
Interests
- Artificial Intelligence
- Computer Vision
- Deep Learning
- Deepfakes Detection
- Quantization, Pruning and Distillation
- Efficient Architectures
Education
PhD in Computer Science 2022 – On Going
Sapienza University of Rome
MSc in Computer Science 2020 – 2022
Sapienza University of Rome
BSc in Computer Science
Sapienza University of Rome
Skills
Programming +10 Years
Deep Learning +8 Years
Computer Vision +3 Years
Research +2 Years
Experience
Research Fellow
Sapienza University of Rome – Now
Rome Italy
- Conducting research on model quantization, binary neural networks, and deepfake detection.
- Presenting research findings at international conferences and workshops.
- Writing and contributing to scientific publications in peer-reviewed journals.
- Securing research grants to support ongoing investigations.
- Supervising and mentoring students in related research topics.
- Collaborating with academic and industry partners on machine learning and computer vision projects.
Research Scientist
University of Klagenfurt – 2024
Klagenfurt Austria
Conducted research on architectural search and optimization techniques for crossview geolocation. Focused on refining model architectures to enhance accuracy, efficiency, and computational performance.
Research Scientist MBDA on Crossview Geolocation
MBDA – 2023
Rome Italy
Conducted research on real-time, low-power cross-view geolocation using transformer-based models. Focused on training and optimizing performance through ONNX format conversion and model quantization to reduce energy consumption and heat generation while maintaining accuracy.
Machine Learning Specialist for Agricultural Water Management
Soonapse srl – 2019
Rome Italy
Conducted research on data mining and preprocessing techniques for optimizing water usage in agriculture. Analyzed and selected machine learning algorithms to model water demand and resource allocation. Developed and evaluated predictive models to improve irrigation efficiency and sustainability.
Researcher Support – Web Infrastructure Development
INFN National Institute of Nuclear Physics – 2016
Frascati Italy
Developed web-based platforms to support research dissemination and collaboration. Designed and implemented flexible WordPress plugins and PHP-based solutions tailored to the needs of scientists for presenting experimental results and project information. Focused on optimizing usability and accessibility for scientific communication.
