ML Infrastructure Developer at Plus

Yandex Plus is a unified subscription to Yandex services, providing access to music, films, podcasts, books, games, sports, and other content. It's a major ecosystem project where over 35 million subscribers use a variety of features across all ecosystem services every day: they listen to 'My Wave' in Music, receive cashback in Taxi, Eda, Market, and other Yandex services, and watch films on Kinopoisk.

Our team is engaged in the development and implementation of ML into various business mechanics, growing key economic metrics. Our focus is on developing projects using classical ML and DL.

We regularly launch new projects and develop ML technologies in our products. We are looking for a talented developer who will help improve the ML infrastructure and take on resource management for solving our tasks. By joining us, you will have the opportunity to see the direct impact of your developments on the speed of ML integration into production and the runtime inference of LLMs.

You will participate in integrating the latest approaches into recommendations for offer products.

This is an opportunity to work at the intersection of cutting-edge technologies and real benefits for millions of people in one of the largest subscription services in Russia.

What tasks await you

Development of runtime ML-backend on Java Your task is to create and develop a runtime framework for ML model inference, to provide ML engineers with convenient and effective tools.

Development and optimization of ML infrastructure components We run many experiments. The development of an experiment management tool is the key to successful implementations. We value proactivity, so we want your own ideas and design of such a system.

Responsibility for resource management We expect you to help the team understand resource limitations and capabilities. You will help ensure team members have access to available computing power, participate in planning necessary resources, and optimize ML pipelines.

Implementation of LLM in runtime production You will implement LLMs in runtime production to ensure their stable and efficient operation in real-world conditions—under high load, with minimal latency, and while adhering to SLAs.

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

We expect you to have

• Confident knowledge of Java or C++
• Experience working with high-load backends
• Experience working with classical and KV databases
• Familiarity with debugging, profiling tools, continuous integration and deployment systems
• Ability to analyze complex data processing processes
• Understanding of the principles of message brokers
• Familiarity with the principles of MapReduce systems

It will be a plus

• Have administered complex services
• Good knowledge of Linux and deep understanding of networking technologies