Pietro Lesci

I am a PhD student in Computer Science at the University of Cambridge advised by Prof Andreas Vlachos. I have a background in economics and 3+ years experience across research labs, consulting firms, and international institutions training custom models and developing data science solutions. Find my cv here .

I am currently interested in understanding how training data influences a model’s behaviour. To study this question, I draw methods from econometrics. I aim to build models that can autonomously select and craft their own training data to continually acquire new capabilities as needed, thus blurring the line between data curation and learning. My research intersects causal methods, active learning, tokenisation, and pre-training.

My work has been presented at major machine learning conferences such as ICLR, ACL, NAACL, and EMNLP. I received a Best Paper Award at ACL 2024 and funding from the Translated Imminent Research Grant for my research contributions. You can find the complete list of publications and related links on the /research page. I also keep my profile updated: each paper is linked to a collection listing all the relevant artefacts.