How Canva Redesigned their Attribution System

Whenever a user logs onto the UberEats app, Uber needs to sort through tens of thousands of meals to generate recommendations. This has to be done in just a few hundred milliseconds. In order to handle this, Uber uses deep learning with Two Towers neural network models. Two Towers is widely used in the industry with companies like Instagram and Snapchat also using it for their recommendation models. Bo Ling is a Staff Software Engineer in AI/ML at Uber and he’ll be giving a fantastic talk on Two Towers and why the model is so widely used in the industry. He’ll delve into the engineering behind integrating these models into Uber’s search platform and Michelangelo feature store. If you’d like to hear Bo’s talk, check out Index. It’s a free community conference for engineers building search, analytics and AI applications at scale. Other speakers include Reynold Xin, the co-founder of Databricks Sudeep Das, head of ML and AI at DoorDash Shriya Arora, engineering manager for personalization

Whenever a user logs onto the UberEats app, Uber needs to sort through tens of thousands of meals to generate recommendations. This has to be done in just a few hundred milliseconds. In order to handle this, Uber uses deep learning with Two Towers neural network models. Two Towers is widely used in the industry with companies like Instagram and Snapchat also using it for their recommendation models. Bo Ling is a Staff Software Engineer in AI/ML at Uber and he’ll be giving a fantastic talk on Two Towers and why the model is so widely used in the industry. He’ll delve into the engineering behind