Backend Developer for the Unified Advertising Database

About us

Advertising is one of the most high-load services at Yandex. It has a runtime that withstands hundreds of thousands of requests per second, and real-time processing that handles tens of gigabytes of data and millions of events per second. The Advertising ecosystem includes many services: from data storage and transformation to ML services for training ranking models. Many of these services need access to advertising data. This is handled by the Unified Advertising Database: it stores data, provides access to it, and transmits changes to related services via a universal protocol. The database contains tens of billions of objects with a volume of over a hundred terabytes.

The Unified Advertising Database is a next-generation database built on YTsaurus dynamic tables. It allows a flexible approach to the data model, using ideas from traditional relational databases and extending them. From an infrastructure perspective, the database combines the capabilities of YTsaurus dynamic tables (transactionality, scalability, fault tolerance) and stream processing. This allows us to combine transactional and non-transactional approaches when solving problems.

Over several years, the Unified Advertising Database has become the center of the Advertising architecture. Dozens of developers from other teams turn to us to solve their product and infrastructure tasks. We create a platform for their solutions and actively implement the database ourselves. We are looking for a colleague who will help with all our tasks, but integration of the database with Yandex Direct is particularly important right now.

What tasks await you

Implementing the Unified Database in Yandex Direct Yandex Direct is one of the largest product services at Yandex, which allows users to launch advertising campaigns and provides a wide selection of advertising inventory. You will continue implementing the Unified Advertising Database in the backend of Direct. This is not just replacing one database with another; it's a serious challenge: you will need to rethink approaches to backend development and transform the service architecture.

Developing the DB core You will develop the database core so that it includes modern functionality and unique capabilities for effectively solving product tasks.

Processing platform In the process of shifting the backend development paradigm, a huge amount of business logic is being moved to real-time processing. You will develop your own framework for streaming data processing to efficiently and conveniently solve product tasks in dozens of processing pipelines.

Infrastructure You bear a great responsibility: to ensure the stability of infrastructure operation, database efficiency, and development convenience, and to ensure that new implementations by colleagues reach users faster. Maximum attention is paid to this.

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

We expect that you

• Develop in different languages: C++ in the database core and processing, Java/Kotlin in the Direct backend, Python — for infrastructure orchestration
• Understand key aspects of building fault-tolerant distributed systems
• Love long-term sustainable solutions but can devise workarounds if necessary
• Are not afraid to read and refactor old code and old solutions
• Write easily readable code

Will be a plus

• Familiar with YTsaurus and stream processing frameworks
• Have production experience with product development in Java/Kotlin
• Have experience working with advertising or advertising-related services, understand the advertising domain model