RocketHuntSmart job search
Reach out directly about this role
By city
3-5 years
Experience
Full-time
Employment
Hybrid, Onsite
Work Format
Backend
Specialization
IT & Tech
Industry
Corporation
Company Type
Backend Developer for Search Projects in Advertising
Advertising is a powerful tool that helps businesses find new customers, enables internet platforms to remain free for users, and allows users to see the most relevant ads for them.
Dozens of large-scale services operate on the search segment of Advertising, processing hundreds of thousands of PRS. Among them are services for candidate generation, model inference, the ad engine with an auction for ads, and others. You will work with complex infrastructure code, understand the final product, and understand ad formats and their impact on the business's product metrics.
Unified Search Auction You will participate in a large-scale project to unify all the technologies used across different segments of search advertising into a single mechanism for processing queries and selecting ads. This is a complex architectural and engineering challenge that will open opportunities for important product implementations and technical optimizations. Currently, there are three main ways to promote a business on Search: special placement (showing ads in top positions), product gallery (product cards before search results), and dynamic placements (promoting a document within the results). Each has its own limitations on banner types, selection algorithms, and auctions. Each segment uses shared and unique data, models, ML factors, features, and embeds. The unified auction will optimize resources: instead of three requests, there will be one, which will reduce the number of calls to external services and computations. Furthermore, the project is important from a product standpoint—the unified auction will provide each segment with information about the others, allowing control over a business's presence across multiple products and coordinating ranking, choosing the best segment for each advertiser.
Development of Dynamic Placements on Search You will develop algorithms for matching documents on the largest document base and advertising banners on the RuNet. Work on auctions using dynamic placements, identify bottlenecks in cross-service ad requests, and collaborate with analysts and the ML engineering team. Dynamic placements on Search are additional ad slots in search results that allow for more clicks and conversions by increasing the visibility of the promoted page without changing the total number of results or their relevance. For the product to work, it is crucial to link the search's document base with the advertisers' banner database and develop algorithms that select thousands of candidates from billions of ads. Developing dynamic placements will help advertisers allocate their budget more effectively and integrate more natively into search results.
More about Backend at Yandex — in the channel Yandex for Backend