ML Developer for the Rhythm Feed Team

We develop, implement, and improve recommendation systems for various Yandex projects with hundreds of millions of DAU and thousands of RPS, processing terabytes of data in real time.

One such project is Rhythm, a commercial content recommendation service integrated across various surfaces, including the Yandex homepage. One of the service's goals is to find and show each user the most relevant content among the available offers.

What we offer:

• The opportunity to improve the service and see the results of your work
• A team with a strong technical background
• The chance to implement your ideas
• Advanced infrastructure for rapid deployment and improvement of recommendations

Join the team whose work influences the choices of millions of people every day!

What tasks await you

Full cycle ML development from idea to implementation Our product and ML development are continuous: we constantly generate hypotheses, test them on data, and implement the best solutions in the service. You will participate in all stages: from research and prototyping of new models to their optimization, production implementation, A/B testing, and subsequent monitoring. The key goal is not only to create a working model but also to ensure its stable, scalable, and efficient operation for millions of users.

Experiments with data analysis, user behavior patterns, and various recommendation schemes The quality of the service directly depends on the depth of understanding of users and data. You will analyze big data, identify hidden behavior patterns, and based on them, explore new approaches to recommendations—from changing candidate generation methods to revising ranking logics.

Training and implementation of generative models Generative models open up new possibilities for personalization. You will research, train such models, and integrate them into the working pipeline of the recommendation service.

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

We expect you to

• Deeply understand the principles of ML and be ready to apply this knowledge to solving real-world problems
• Know C++ or want to learn it
• Have a good command of Python
• Have a genuine interest in implementing changes that significantly impact user services

It will be a plus if you

• Have experience with recommendation systems or NLP
• Are a C++ expert
• Have worked with transformer models
• Have worked with PyTorch or other deep learning frameworks