ML Developer for the Detector Alignment Team (Autonomous Transportation)

The Detector Alignment Team within the Autonomous Transportation Perception Department is responsible for ensuring the car correctly "sees" and interprets the surrounding environment under any operating conditions: day and night, in rain, snow, and fog.

Our task is to ensure the coordinated and stable operation of detectors and segmentation models in the real world. We work with camera, lidar, and radar data, create neural network algorithms, and bring them to reliable production on the vehicle.

What tasks await you

Development and improvement of detection and segmentation models Training and adapting models for the detection of cars, pedestrians, and other dynamic objects, segmentation of the roadway, and highlighting important scene elements.

Alignment and stability of models in real-world conditions Working on the generalization ability of models in various weather and lighting conditions, on different sensor configurations, and in new geographies.

Multimodal data processing Integrating information from cameras, lidars, and radars, and exploring ways to combine it effectively.

Determining acceptable operating conditions Developing algorithms to determine weather conditions, road state, and other factors affecting the ability to drive safely.

Production and scaling Optimizing models, analyzing errors, working with large volumes of data and training infrastructure.

More about ML at Yandex — in the channel Yandex for ML

We expect you to

• Train neural network models, work with different approaches
• Have worked with PyTorch or TensorFlow and frameworks built on top of them
• Strive to use and spread universal, scalable practices in your work
• Be familiar with the world of applied DL: know computer vision theory, achieve practical results
• Understand the specifics of real-time constraints
• Read papers from ML conferences, follow the development of the field

Would be a plus

• Have worked with multimodal models (camera + lidar + radar)
• Have experience with large models (LLM, VLM, Foundation Models) in other domains
• Have performed autonomous transportation tasks