I am Pietro Lesci, a PhD student in Computer Science at Cambridge University working on Machine Learning for Natural Language Processing.
Prior to starting the PhD in October 2021, I have been a senior associate in data science at Bain & Company for 1.5 years. I have a background in economics (both BSc and MSc) and I slowly transitioned to data science and eventually to research in machine learning (ML) and natural language processing (NLP).
I started my journey from economics to ML in 2018 while working as a trainee in data science at the Directorate of General Statistics at the European Central Bank in Frankfurt am Main (Germany). During this period, I got formal recognition as being the only trainee to have ever developed a new procedure end-to-end, coordinated the expert counterparties from the National Central Banks during its implementation, and presented during regular plenary meetings. Given my performance, I received a contract extension.
Once back from Frankfurt, I graduate with an MSc in Economic and Social Sciences at Bocconi University in 2019. The course was centered on Bayesian methods for computational social sciences and featured many applied econometrics and time-series courses. I wrote my thesis under the supervision of Prof Sonia Petrone. In the thesis, I discuss the statistical interpretation of a new class of algorithms called Neural Processes proposed that very same year.
After graduation, I properly entered the ML/NLP world by working as a research assistant at the Bocconi Institute for Data Science and Analytics with Prof Dirk Hovy in Milan (Italy). I was a member of the team working to understand, explore, and measure the health of conversations on the Twitter platform - the project was completely funded by Twitter. As part of my work at Bocconi University, I deployed two machine learning web apps — Wordify and MACE — that allow researchers in other fields (e.g., social sciences, marketing, economics, etc) to easily access NLP tools.
From March 2020 until the start of the PhD, I have been working as a data scientist in the Advanced Analytics Group at Bain & Company, a strategic consulting firm. There, I had the chance to work on many exciting ML/MLOps problems across a variety of industries (telco, finance, consumers goods, mining), continents (Europe, America, Asia) programming languages (Python, R, Julia, SQL, Rust), and use-cases (analyses, software projects, libraries, products, and pipelines). Being Bain & Company one of the top 3 strategic consulting firms in the world, I had the great opportunity to learn the “consultant curriculum”: managing teams, scoping work, and dealing with tight deadlines while attaining the highest required quality standards for our clients. I am glad I had the chance to be managed by Dr Diane Berry and Marc Van Heerden during the time spent there.
I believe this is my personal added value: I worked in very diverse environments collaborating with people at different levels of seniority and from different cultures. I always tried to humbly give my best and often proved to be able to fast-learn, adapt, and succeed even in contexts that are not my own. Most importantly, I luckily managed to always leave with more friends than when I arrived.
As of October 2021, I am a PhD student in Computer Science working on Machine Learning for Natural Language Processing with Prof Andreas Vlachos. I am currently interested in parameter- and data-efficient fine-tuning of large language models (keywords: few-shot learning, meta-learning, active-learning). Also, I like to contribute to open-source software projects. I have recently become a core maintainer of the lightning-flash Python library.
First of six children, I am a passionate musician and a statistics enthusiast.