Yandex ORD (Advertising Data Operator) is a large-scale infrastructure designed for participants in the online advertising market and enabling compliance with the advertising labeling law. Our service provides verification, reliable data storage, and data transfer to the Unified Register of Online Advertising (ERIR).

Our project's goal is to create a convenient tool for both Yandex's internal advertising systems and external agencies, publishers, and other participants in the advertising market. The ORD project is only one year old — the codebase is new, and there is no legacy code.

What tasks await you

Development of data processing functionality Our service is at the center of a rapidly developing advertising technology market. Infrastructure investments are particularly important here. At ORD, you will have the opportunity to work with Yandex's cutting-edge streaming data processing solutions. We plan to add new processing pipelines to the event-driven service architecture.

What you will do: * Develop a deduplication pipeline for objects coming from different systems and users * Work with product teams to find labeling solutions for new advertising mechanics and products * Prepare and defend your solutions at architectural reviews with development and analytics teams

Development of ORD as a product At the start of the project, the priority for Yandex ORD was labeling data for the Yandex Direct advertising system, which generates over 200 million requests per day just for labeling creatives. Now we aim to make ORD more accessible and convenient for a wide range of users.

What you will do: * Work with the UI team to develop the backend for the user interface * Analyze client difficulties and simplify the entry and correction of data

Optimization and scaling of data processing pipelines Labeling is mandatory for all of Yandex's advertising infrastructure, so service stability is critical. Every day we receive, process, and send hundreds of millions of objects. It's important to ensure high horizontal scalability in all parts of the system.

What you will do: * Develop monitoring systems, implement and improve incident management practices * Develop the system for sharding data sources. Objects related by business logic must be processed by one process, and in the correct order. At the same time, maximum parallelism must be achieved to minimize delays * Identify bottlenecks in processing and remove them from the object's critical path. Example: moving the processing of video and image creatives to a separate pipeline * Automate the movement of data from hot to cold storage. ORD stores data for years, while users need a small subset of active documents — storing everything with the same accessibility would be inefficient

We expect you to have

• Familiarity with classic algorithms and data structures
• Attention to detail and meticulousness
• Excellent knowledge of C++ and STL
• Proficiency in SQL query language
• Experience with relational and non-relational databases

Will be a plus

• Experience developing high-load distributed systems
• Familiarity with queue systems (Kafka, RabbitMQ)
• Strive for optimal and concise code
• Experience in Linux OS and network administration