Mírian Silva

Research Interest

• In addition to my engineering background, I have research interests in topics such as Fairness and Bias in AI, as well as ethics and trustworthy AI. Interdisciplinary research in AI and Law, AI and Social Sciences, and AI and Public Policy is also part of my research interest. If you are working on these or related topics and are seeking research collaborations, please feel free to reach out to me via email or social media.

TLDR: Academic and Professional background

• I've been an AI Engineer at IBM Research since 2021. I work with the AI Models Team that works alongside the MIT-IBM Watson AI Lab, mainly doing collaborative research that relates to generative AI. Currently topics of work include: LLMs evaluations, validation process, prompt engineering and AI agents assistants.
  • I've been collaborating closely with product managers and researchers to evaluate models, agent systems, and algorithms that are continuously adapting and improving based on users/clients interactions and feedback, aimed to drive adoption of proprietary models.
  • Open-source projects: github.com/IBM/ibm-generative-ai. A Python library built on IBM's large language model REST interface to seamlessly integrate and extend this service in Python programs.
  • Research publications from my work, can be found in my page at the IBM Research profile.
• I recently received a Master's Degree in Computer Science at the Federal University of Minas Gerais (UFMG, Brazil), with research based on AI focused on machine learning fairness, bias mitigations, and evaluation practices. I also hold a B.Sc in Computational Mathematics, a degree based on computer science and applied mathematics, from the same university (UFMG).
  • Mírian Silva. (2024). A study of the nuances of ai fairness development in practice: A framework for designing bias mitigation interventions (Master’s Thesis, Federal University of Minas Gerais (UFMG), Belo Horizonte, Brazil). Available at http://hdl.handle.net/1843/78989.
• Previous Research Fellow and Visiting Scholar at the University of California, Berkeley, in the Berkeley Equity and Access in Algorithms, Mechanisms, and Optimization (BEAAMO) Research Group. Working on auditing and evaluating algorithmic tools and evidentiary statistical software used in criminal legal systems. Supervised by Prof. Rediet Abebe, Ph.D.
• I usually organize workshops at top-tier AI conferences such as NeurIPS, ICML, and NAACL, as chair organizer of Black in AI affinity groups.