Recent Ph.D. graduate in computer science with five years of experience publishing in top artificial intelligence, machine learning, and data science venues. My research is aimed at engineering trustworthy AI systems in fields where the cost of making a mistake is high. Because the fields of AI & ML are constantly evolving, I have had the chance to work with the most impactful algorithms being used today, including large language models (transformers), computer vision systems, natural language processing, anomaly detection, and more. Some examples of my research include:Cooperative Knowledge Distillation: A Learner Agnostic Approach (Published in AAAI24): We propose a system in which classifiers encode learned information into virtual data and share that data with other classifiers to improve performance - all while maintaining the privacy of the original training data.Foundations for Unfairness in Anomaly Detection (Published in AAAI/ACM AIES24): One of the predominant theories as to why machines are unfair is that unfairness is encoded into the label space, but this does not explain why unsupervised learners are also observed to be unfair. We propose a model for how machine bias can occur in such unsupervised settings, and demonstrate that this model provides a clear and full picture as to why such models exhibit bias.Explainable Clustering via Exemplars: Complexity and Efficient Approximation Algorithms (Published in SIAM SDM24): We propose a system of simultaneous clustering and explanation where the explanation is not grounded in the input features, but rather in particularly important instances which we call exemplars. This is particularly relevant in 2024, as increasingly the input space is not inherently interpretable, such as embeddings from transformers.I am excited to continue my work in ensuring that AI is trustworthy and reliable. If you would like to connect, feel free to connect with me here or over email at MJLivanos@UCDavis.edu.