CV

Find the complete CV by clicking on the PDF icon on the right.

General Information

Full Name Riccardo Majellaro
Email riccardomajellaro [at] gmail [dot] com

Education

  • sep 2021 - jul 2023
    MSc Computer Science (Artificial Intelligence track)
    Leiden University, Leiden, Netherlands
    • GPA: 8.9/10 Cum Laude (with honors)
    • Thesis: "Disentangling Shape and Texture Dimensions in Object-Centric Representations" (paper)
    • Courses
      • Introduction to Machine Learning
      • Statistical Learning
      • Introduction to Deep Learning
      • Advances in Deep Learning
      • Reinforcement Learning
      • Seminar Advanced Deep Reinforcement Learning
      • Text Mining
      • Advances in Data Mining
      • Automated Machine Learning
      • Multicriteria Optimization and Decision Analysis
      • Evolutionary Algorithms
      • Robotics
      • Modern Game AI Algorithms
  • sep 2017 - oct 2020
    BSc Computer Engineering
    University of Modena and Reggio Emilia, Modena, Italy
    • GPA: 8.8/10
    • Thesis: "Distributed Training of DETR on Marconi100"
    • Courses
      • Calculus I
      • Calculus II
      • Applied Mathematics and Statistics
      • Geometry and Linear Algebra
      • Operative Research
      • General Physics
      • Applied Physics for Informatics
      • Control systems
      • Fundamentals of Computer Science I
      • Fundamentals of Computer Science II
      • Operating Systems
      • Computer Architecture
      • Object-Oriented Programming
      • Databases
      • Computer Networks
      • Software Engineering
      • Dynamic Languages
      • Digital Electronics
      • Fundamentals of Telecommunications
      • Introduction to Business Management

Experience

  • jun 2024 - present
    Machine Learning Engineer
    DuckDuckGoose
    • Research on deepfake detection and explainability methods.
    • Designed and conducted experiments to improve internal and academic SOTA deepfake detection models, achieving up to 15% accuracy improvement on fake samples at very low false positive rate (~0.5%) while reducing model size 6.5x and inference speed 2.5x.
    • Led research and development of novel explainability method for deepfake detection models, outperforming current academic SOTA.
    • Wrote production-level code to deploy ML models, optimizing speed and memory requirements.
    • Designed algorithm to minimize models deployment costs while maintaining efficiency by leveraging clients usage statistics.
  • feb 2023 - jul 2023
    Teaching Assistant
    LIACS, Leiden University
    • Teaching assistant for the MSc course "Reinforcement Learning" taught by Prof. Aske Plaat
    • Correcting and grading assignments. Helping students via email and during workgroup sessions

Languages

  • Italian (native), English (fluent)