Every day, millions of people use Yandex Maps to navigate or to get from point A to point B on foot, by car, or by public transport. To keep the map up-to-date, we need to monitor changes in the real world, collect reliable on-the-ground signals, interpret them, and ultimately update the map.

Changes occur for various reasons:

• road services have systematically changed the traffic pattern and installed a road sign or changed the road surface markings;
• traffic on a bridge has been blocked due to an accident;
• residents of a building have installed a barrier at the entrance to the courtyard;
• someone has built a new house;
• a sign on a building facade has changed because a new cafe has opened.

The key goal of our team is to reduce the time between changes in the real world and their representation on the map. To achieve this, we use ML to detect changes via GPS tracks, detect road infrastructure objects in photos, interpret the collected data, and apply changes to the map.

We have a strong team that has already automated several continuously running processes. For example, we have created a process for adding new buildings to the map using satellite imagery.

What tasks await you

Developing the system for detecting changes in road infrastructure You will develop a system that helps identify the appearance and disappearance of road signs, changes in road markings, and the opening/closing of roads using photos, GPS tracks, and internet messages.

Developing a system for automatic detection of errors on the map You will need to develop a system for validating the consistency of map data and ground-truth sources (photos, satellite imagery).

Creating a copilot mode in the map editor You will create mechanisms to speed up making edits to the map through automatic context summarization and smart prompts.

More about backend at Yandex — in the channel Yandex for Backend

We expect you

• Know C++, Python
• Have a good understanding of machine learning principles and have applied them in practice