Uber’s Hyderabad-based Risk Engineering team has increased efficiency of the platform to 99.999% from 99.9% two years ago. The improvement of the efficiency by two decimal points translates to a hundred-fold increase of the security on the platform, thereby providing a safe ecosystem to millions of people transacting on it by the day.
The team did this by using state-of-the-art, ML-based compute virtualization and optimization techniques, minimising the memory and CPU overhead for every request. Alongside, they isolated heavy input-output processing needs, ensuring optimal performance for billions of calls per day. The various tools deployed work in conjunction to detect and prevent unauthorised behavior within a fraction of a second, thereby enhancing the already seamless experience for riders, eaters, driver partners and courier partners.
Commenting on the increased efficiency, Naga Kasu, Director – Engineering, Uber, said, “It is our top priority to work towards making the platform as safe as possible. We understand the trust all our stakeholders put in us everyday, and strive tirelessly to ensure their financial information is secure and that they transact on the platform with surety. The improvement in our system architecture is a fine example of the results the team delivered.”
The Risk Engineering team is responsible for all transactions on the platform, including those by riders, eaters, earners, and the business, all of whom depend on it to keep the system safe and secure.
The team is a group of 15 agile engineers, proficient in languages such as Java, Go, JavaScript, React and skilled in handling extremely large scale distributed systems, designing flexible control planes and building massive data pipelines for intelligent inferencing systems.
