Lidar Map Infrastructure Developer

Lidar Map Infrastructure Developer

Autonomous vehicles must always know their exact location in space. One of the most reliable methods for solving this task is localization using a lidar map: a three-dimensional reconstruction of the world in the form of a point cloud.

The Lidar Map Infrastructure development team builds scalable systems for the automated creation and updating of maps, and also creates engineering solutions to optimize the processing of large volumes of geospatial data.

What tasks await you

Scaling and optimizing the map-building platform We need to cover tens of thousands of kilometers with a lidar map. To do this, we are learning to build maps very quickly, which requires refactoring the system architecture, scaling data processing pipelines, and monitoring the quality of all systems. We are working on intelligent orchestration and batching of build operations to gain speed.

Designing a system for monitoring map relevance and rebuilding outdated maps We are developing tools to automatically detect map obsolescence, orchestrate rebuild processes, and manage versioning of geospatial data in real time.

Developing high-load point cloud processing systems Building each kilometer is a task of point cloud optimization, requiring numerical optimization and repeated solving of systems of linear equations. We optimize memory and CPU-critical operations, solve caching tasks, and manage resources when processing complex 3D structures.

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

We expect that you

• Have deep knowledge of C++ and Python
• Have processed big data in distributed systems
• Understand the principles of scalable system architecture, know how to work with memory and bandwidth constraints
• Are able to profile and optimize high-load code

Will be a plus

• Have worked with cloud data orchestration systems (Yandex Nirvana, Airflow, Dagster, or equivalents)
• Have solved engineering tasks involving graph optimization (SLAM, factor graphs)
• Are familiar with point cloud registration algorithms (ICP, NDT)
• Are familiar with geospatial data or cartography systems