Uber has an in-house search engine called Search In Action (SIA). As the backbone behind the feed and search capabilities of Uber's Delivery business, SIA plays a crucial role in expanding selection seamlessly for customers which is a strategic advantage to the business.

SIA has continually evolved to enable customers to find their favorite stores, even if they're not conveniently located for pickup. Join us as we delve into the real-world challenges faced in handling millions of merchants with dynamic traffic patterns, all while striking a balance between backend architecture performance and incremental order volume.

In this session, we'll dive into the heart of the matter- the optimization problem of the query layer with a dive deep into how internals of SIA works, challenges faced and how we built a solution which is scalable for multiple similar use cases.

Learn firsthand how we tackled this challenge by devising a novel index layout specifically tailored for range queries. The result is an impressive 40% reduction in latency without impacting the business's bottom line.

Gain invaluable insights into building cost-effective in-house solutions in today's cloud-dominated landscape as we share essential tips on avoiding pitfalls while working on large scale systems and achieving incremental optimization wins.

Speaker

Janani Narayanan

Applied ML Engineer @Uber, Previously Tech Lead on DynamoDB Control Plane (Early Stage), 10+ Years Tech Industry Experience

Janani is a Senior Staff Engineer and Technical Lead Manager at Uber Technologies, where she has been at the forefront of transformative zero-to-one (0-1) initiatives for the past nine years. Prior to her current role, she was a pivotal contributor to SimpleDB and DynamoDB, playing a key role in their development during the initial four years post-public launch. Her work has led to several patents across both DynamoDB and Uber.

At Uber Eats, Janani leads a specialized "tiger team" focused on addressing the cold start problem by understanding selection. Her team's mission is to solve challenges related to eater and merchant retention by improving the precision of machine learning models that drive eater targeting, merchant acquisition, growth, personalization, ranking, and recommendations. Previously, she has successfully scaled UberEats’s Ads, Feed, and Search backends by 3x while maintaining low latency, optimized global rider pricing through advanced ML pipelines, and developed precise ETA predictions for accurate upfront pricing.

Janani has also built foundational infrastructure for anomaly detection in fares, created targeted rider promotions powered by machine learning, and developed distributed event processing systems capable of handling trip-scale data. These systems are crucial for real-time and batch processing, underpinning Uber’s real-time machine learning predictions across its marketplace, finance, and accounting functions.

Speaker

Karthik Ramasamy

Senior Staff Software Engineer @Uber, 15 Years of Experience in Design and Implementation of Web Applications, Distributed Systems, Search and Analytics Infrastructure

Karthik Ramasamy is a Sr Staff Software Engineer at Uber. He is currently focused on building distributed search platforms to power various Uber's search use cases.