Opportunities and Challenges in Building a Machine Learning-Centric Company
On May 23 the Stanford Club of France was invited by Criteo for an evening of talks about Machine learning, followed by a tour of the Criteo AI Lab and a cocktail on the Criteo panoramic rooftop terrace.
In 2005, Criteo has developed into a $2.5B global business by competing against companies such as Google, Facebook, or Amazon. The “secret sauce” of Criteo is its Engine, an application of various machine learning techniques that have been carefully optimized over time. Last year, they opened their AI Lab, which currently comprises over 80 world-class researchers and engineers.
In this presentation, we discussed what has prompted the need for this considerable investment in machine learning, as well as some of Criteos success stories and pitfalls. We argue that the state of the art of machine learning, as reflected in our experience, may have profound implications on the dynamics of whole industries. The investments required risk dividing players into “haves” and “have-nots”, something that may disrupt whole industries in the coming years and decades.
- Intro and overview of machine learning at Criteo (Dan Teodosiu)
- The difference between theory and practice: a brief history of the Engine (Romain Lerallut)
- State of the art in machine learning and what’s next (Flavian Vasile)
- Q&A (all)
Dan Teodosiu - Chief Technology Officer
Dan has extensive leadership experience in both startups and large companies, as well as an entrepreneur in Silicon Valley, Seattle, and Europe. He’s passionate about using technology to solve challenging business problems, building world-class engineering and research teams, and creating a strong, empowerment-based engineering culture. Dan joined Criteo in 2013 and leads all engineering, research and production. He has successfully scaled Criteo’s world class R&D team to over 600 engineers, enabling the company’s successful IPO in 2013 and setting the stage for long-term growth. Prior to Criteo, he worked at Google, Microsoft and Hewlett-Packard, and co-founded two companies in the Bay Area, one of which was acquired by Microsoft. Dan holds an MS (’96) and PhD (’00) in Computer Science from Stanford University, and is an inventor on over 50 patents.
Romain Lerallut - Engineering Director
Romain Lerallut leads the Engineering branch of the Criteo AI Lab that is developing the Machine Learning platform supporting all Research and Product uses of AI at Criteo. Before the launch of the Lab in 2018, he spent 6 years as a manager in the Engine department, in charge of building large-scale machine learning algorithms to solve actual problems such as product recommendation or graphical layout optimization. Before joining Criteo, he was teaching computers how to read cursive handwriting at A2iA, a world-leading company in this field. Romain holds an engineering degree from Ecole des Ponts-Paristech and a PhD in Computer Science from Ecoles des Mines-Paristech.
Flavian Vasile - Research Lead
Flavian Vasile is a Research Lead in the Criteo AI Lab, where he works on the development of Deep Learning-based Recommendation Systems. Before joining Criteo, he worked as a Senior Researcher in the Twitter Advertising Science team; before that, in the Yahoo! Research Lab where he mostly focused on Content Understanding problems. His current research interests include Deep Sequential Models for Recommendation and understanding Recommendation as a decision-making system with reward uncertainty. Among his recent research publications, the work on Causal Embeddings for Recommendation received the best paper award at RecSys 2018 and he is the co-organizer of the Workshop on Offline Evaluation for Recommender Systems. Flavian holds an MSc degree in Machine Learning from Iowa State University.