Inna Dvortsova

Senior Software Engineer

Inna Dvortsova graduated from Taras Shevchenko National University of Kyiv with a masters in mathematics in 2013. She worked as a practicing software engineer at several companies in the IT sector before moving to take up a post as tech lead at Barclays in 2015, where she led data analysis, algorithms and deployment of stress testing. In 2018 she moved to Spotify, where she developed and deployed systems that supported and optimised the company’s revenue streams. More recently in 2020 she started work at Facebook where she has been a lead engineer on the technical development, deployment and optimisation of Facebook’s WW Web Enabled Simulation system. She and her co-authors have published results from this work at the 8th Genetic Improvement Workshop (GI 2020) and the 43rd International Conference on Software Engineering (ICSE 2021). She has forthcoming keynote papers at the 9th International workshop on Realising Artificial Intelligence synergies in Software Engineering (RAISE 2021), and the 15th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM 2021) and she recently gave the keynote talk at the 25th International Conference on Evaluation and Assessment in Software Engineering (EASE 2021).


Beyond end-to-end testing: a new approach to system testing using Facebook cyber-cyber digital twin

Is end to end testing enough? The end to end testing we know and love tends to test a single user interaction with the system. What happens when multiple users interact with one another on a software platform? This is increasingly the scenario faced by software testers and is not well-served by current approaches to end-to-end testing. In this keynote, I will describe how Facebook is tackling these testing challenges using simulation-based testing. We have built a cyber-cyber digital twin of Facebook’s entire backend systems (of hundreds of millions of lines of code). The system is called WW, and it allows us to test user interaction scenarios at scale. WW is a controlled environment for testing in which bots act as users, but all interactions occur on the real platform infrastructure. In the talk, I will outline WW’s development and deployment and how they are using it to catch bugs at Facebook.