Complex Systems Summer School
Educated in several acronyms across the globe (UNISR, SFI, MIT), Jacopo Tagliabue is now the Lead A.I. Scientist at Coveo, after the acquisition of his own Artificial Intelligence (A.I.) startup Tooso in 2019. He works on theoretical and practical challenges in A.I., shipping Machine Learning models to hundreds of millions of devices every year. In previous lives, he completed a PhD, worked for a professional basketball team, and gave an academic talk on video games (among others improbable "achievements"); his work has been often featured in the general press and presented internationally in scientific and business venues. He attended Complex Systems Summer School in 2009.
Briefly describe your primary research/academic work or other professional work.
I am currently working as a Lead A.I. Scientist at Coveo, a North American company providing Information Retrieval and Recommendation services to a network of more than 500 companies. Together with my team, we do theoretical and applied research in A.I. and machine learning (as a non-exhaustive list of recent topics, personalization, information retrieval, causal inference, time-series classification, recommender systems).
Building effective industrial-scale machine learning systems is a multi-disciplinary (complex, you could say!) effort, but after all these years my favorite part is still natural language processing (NLP). I have a lifelong interest in (natural and formal) languages and my previous A.I. company - Tooso - operated around the idea that our interaction with search engines should be closer to talking to a child than a salad of keywords.
In what ways does the study of complexity science influence your thinking about your current work?
As an entrepreneur, the scaling laws integral to the functionality of complex systems have very practical implications. Building companies with good scaling properties is not an endeavor born out of theoretical curiosity. Those properties are actually critical to engineering a culture, a business model and a technology roadmap that grow together in a healthy way.
As a researcher, the entire SFI toolkit is always with me, ready to be used when I see an opportunity. For example, in a recent work on the emergence of communication, we were able to push the debate a tiny bit further by introducing some graph thinking - i.e. the idea that intelligent agents don’t live in a vacuum - they actually have underlying social connections whose topology subtly influences global outcomes.
How did your experience at CSSS impact your professional (or personal) perspective?
I have always been more of a generalist than a specialist, and science as practiced at SFI made a lasting impact on how I approach research questions. SFI emanates a multi-disciplinary, creative and serendipitous vibe that I simply haven’t found elsewhere.
Finally, it may sound a bit cliche, but it is very true that at SFI the “best things are not things”. I have fond personal memories of that summer, especially given that I made lifelong friends. Of all the papers we’ve recently published, the one on Nat. Sci. Rep. is special, as it was designed together with Lucas - a friend from CSSS.
What interests do you have that might surprise your colleagues?
I have taken on every possible role in basketball: player, fan, coach, referee and… scientist! As the legend goes, quite some time before Moneyball was a Brad Pitt movie, I was involved in the design of one of the first systems to analyze basketball data for a professional team (the mythical Olimpia Milano!). The software was horrible by today’s standards (while I still suck at coding at 35, it was a whole different level of sucking when I was 23) and pretty simple (some descriptive stats and few predictions), but it was - in its own way - a completely new way to look at a complex sport. If you flash-forward 10 years to today, all major teams in the NBA and Europe have made significant investments in sport analytics. Given my love for The Game, I am happy and proud to have played a “walk-on part” in that war.
This interview was conducted in October of 2020