About Me

Hi there! 👋 I'm Arthur Dantas Mangussi, a Machine Learning researcher with a passion for developing innovative solutions in Data-Centric AI, with a particular focus on Missing Data. My research spans multiple intersections, including missing data imputation, its relationship with noisy data, fairness, and adversarial machine learning.

I'm also deeply fascinated by Large Language Models (LLMs) and enjoy exploring how cutting-edge technologies can effectively address real-world challenges.

🎓 Academic Background

• Master's Degree in Operations Research and Data Science

  • Institution: Aeronautics Institute of Technology (ITA) and Federal University of São Paulo (UNIFESP), Brazil
  • Research: Focused on Data-Centric AI, exploring challenges related to missing data and other real-world data quality issues, including noise and fairness.

• Master Internship at the University of Coimbra (UC), Portugal

  • Explored the use of Autoencoders (AEs) for Missing Data Imputation. Additionally, I began coding with a focus on prioritizing parallelization and optimizing methods for computational efficiency.
  • During my stay at the University of Coimbra (UC), I developed a Python library called mdatagen, designed to simulate artificial missing data scenarios. The library is publicly available on PyPI.
  • I also worked on improving my technical English, particularly for academic writing and professional conversations. My current level is CEFR B2, with a Duolingo English Test score of 110.

• Bachelor's Degree in Medical Physics

  • Institution: Federal University of Health Sciences of Porto Alegre (UFCSPA)
  • Achievements: Developed the AQMI software, a tool to assess the quality of mammography images. The codebase is available on GitHub. The original paper was published in the Brazilian Journal of Radiation Sciences

💻 Technologies & Tools

Here are the technologies I work with most frequently:

Programming Languages

Libraries & Frameworks

• Machine Learning & Deep Learning: TensorFlow, scikit-learn
• Data Analysis: pandas, NumPy, matplotlib, seaborn
• Fairness & Bias Mitigation: AI Fairness 360, Fairlearn
• Adversarial Attacks: ART (Adversarial Robustness Toolbox)

Tools & Platforms

• Development: Jupyter Notebook, VSCode
• Scientific Writing: Overleaf, LaTeX
• Version Control: GitHub

📫 How to Reach Me