Date / Heure
Date(s) - 15/10/2019
18h30 - 22h00
Emplacement
Criteo
Catégories
With an ever increasing number and diversity of Digital Services, fraud has become a hot topic in all its various forms. In this context, Criteo Labs is excited to bring you our first Digital Fraud meetup. We will exchange our best practices to defeat fraudsters, through a great line-up of talks from both our Criteo R&D Engineers and our guest speakers.
⌚ Agenda :
6:30 pm : welcome
7:00 pm : talks
9:00 pm : networking
_______________________________________________________________________________________
🎤 Cécile Lefèvre-Ardant – Data Scientist product analytic, Michel Ying – Product manager & Fares Sabbagh – developer at Criteo
📃 Why & how Criteo deals with invalid traffic
Advertising fraud is a rampant and growing issue in the programmatic industry. In practice, Criteo has to analyze each of the 300B requests we receive per day in order to detect suspicious traffic.
We will deep dive into the solutions we put in place to tackle these challenges.
🎤 Pierre Houssin – Data Scientist of Adevinta
📃 Content moderation at Adevinta
During the talk, I will explain how automatic moderation has evolved at leboncoin, and more generally at Adevinta. From a monolithic & single tenant to a multi-tenant architecture (running on AWS), I will present you what was put in place in terms of architecture, infrastruture, data pipeline, as well as the challenges we faced.
🎤 Pierre-François Bouquet – Data scientist at Blablacar
📃 Frontend navigation behaviour analysis for fraud prevention
Fraudsters are behaving differently, it shows up into activity metrics, volumes, types of transaction, etc … To perform on apps breach, scale their activity and spend less time putting it in place, they are manipulating the apps and website way differently than legitimate consumers. We’ll demonstrate how BlaBlaCar is analysing screen sequences to be able to detect fraud activity in minutes after fraud events.
https://www.meetup.com/fr-FR/Criteo-Labs-Tech-Talks/events/264951942/
