Geo Search runs the search that tens of millions of riders use to say where they are going, across many countries where commercial maps are often wrong, incomplete, or missing
The same search and location capabilities also power other verticals such as courier, cargo, and intercity
The team owns place search and autocomplete, geocoding, ranking and personalization, location detection and pickup selection, integrations with external providers such as Google and 2GIS, and its own search screens on Android and iOS
Measured on order conversion rate, search conversion, successful search sessions, and mean reciprocal rank
Responsibilities
Design and build the deep learning systems behind search ranking, session-based recommendations, and multi-objective personalization, learning from rider behavior and serving users across markets
Make ranking consume geographic context and own the recommendation and ranking of pickup points
Translate business goals into ML objectives with non-functional requirements
Lead evaluation end-to-end, from offline metrics to the design of online A/B tests, and prove a change improves engagement before it ships
Partner with backend engineers to take models from prototype to production, including model serving and latency
Partner with the product manager and operations to turn behavior analysis into concrete features and requirements
Own the ML lifecycle in production, serving, monitoring for drift and building the retraining pipelines that hold quality as data shifts