Michael Accetto

I build systems using grounded, interpretable inference for complex domains. With my team I develop responsible AI strategies to deliver expert systems in an ergonomic package. We keep humans in charge, monitor anomalies and design feedback loops for continuous improvement.

I lead SMEs in the design and deployment of AI systems. I focus on scoped projects with real impact to build confidence towards AI driven change within the organization. I often turn complex sets of business needs into technical requirements to implement effective ergonomic systems. I focus on fine-tuning and deployment of complex AI Workflows, enabling domain specific inference by using neurosymbolic methods (GraphRAGs, HybridRAGs). Notable projects include Pansplore, an AI operating system for DMCs, and TTC Maker, a biometric calculator and routine generator.

Fine-tuned LLaMA and Falcon models on proprietary legal data, deploying them via cloud infrastructure (VMs, Docker) to support automated lawsuit drafting for legal professionals. Authored a piece on psychological bias in legal decision-making to bridge AI research and legal practice. Initiated discussions with the management team on company culture, contributing to the launch of internal initiatives focused on cultural improvement.

Taught Courses: - Knowledge and Data (2020). Topics: Symbolic Reasoning, Probabilistic Inference, Ontology Creation. - Intelligent Systems (2020-2021). Topics: Bayesian Probability, Formal Logic, Machine Learning + Search Algorithms. Tutoring: - Knowledge Engineering (2018, 2019). Topics: Knowledge Engineering (commonKADS methodology) - The Social Web (2019, 2020). Topics: Network Analysis, NLP.

VU research fellowship 2018-2019 in the field of Cognitive Sciences.