Research Focus
My principal research interest is making a rigorous real-world impact through the implementation of Artificial Intelligence in medicine — from designing evaluation frameworks for clinical AI agents to decoding brain signals.
Master's Thesis: Evaluating Clinical Agentic AI Systems in Oncology
Multi-tier evaluation framework to validate autonomous AI agents for NSCLC treatment planning. Combines LLM-as-judge, deterministic rule checks, and clinician feedback loops to ensure safety, accuracy, and alignment with clinical guidelines.
Thesis RepositoryResearch Internship — AI-On-Lab
Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy
Multidimensional Evaluation Framework for Clinical AI Agents
Designed and developed a comprehensive, multi-tier evaluation framework to bench-test and validate autonomous AI agents specialized in Non-Small Cell Lung Cancer (NSCLC) treatment planning. Architected a 5-tier metric system covering reasoning, output quality, safety, workflow, and reliability — establishing standardized KPIs for clinical LLM deployment.
This work culminated in a primary-author submission to the ESMO Congress 2026, in close collaboration with cross-functional medical and technical teams.
Research Areas
The problems I care about span the full pipeline of medical AI — from understanding how models reason over clinical data to deploying them safely.
Agentic AI & Evaluation
Designing autonomous agent architectures and rigorous evaluation frameworks (LLM-as-a-judge) for high-stakes clinical decision support.
Digital Pathology
Image classification for benign vs. malignant features, confidence-interval estimation, and full-slice vs. zoomed-slice multi-input pipelines.
fMRI & Neuroengineering
Decoding visual stimulus categories from human brain activation maps with ROI-based analysis and voxel-level explainability.
Biomedical NLP
Conversational AI for medical question-answering, fine-tuned on biomedical datasets such as PubMed Q&A.