C++ Developer for the Distributed Bidder Infrastructure Team in Advertising

We work on the infrastructure of one of the key systems in advertising — the autobudget (also known as the autobidder or auto-strategies).

The autobudget manages all advertising traffic. It is the system that solves a complex optimization problem and determines how many and which ad banners will get impressions, clicks, and conversions to spend advertisers' money as efficiently as possible. All of this is managed by a massive and high-load infrastructure that reads data streams from across the entire internet, calculates numerous neural models, and then assists in ranking and final determination of the winner in each auction per query.

In addition to developing the autobudget as a service, we need to think about advertisers. Every day, millions of clients create new ad campaigns, modify, and update old ones. Here, we implement new ways of managing ad campaigns, come up with new settings, optimization methods, and budget constraints.

The result is a complex, infrastructure-heavy product close to Machine Learning, which directly impacts Yandex's business and revenue.

About the Team

The team is concentrated in Moscow, there is no strict schedule, but we like to go to the office to communicate. The team consists of strong and young guys with degrees from FCS/YSDA (Faculty of Computer Science/Yandex School of Data Analysis). Besides work, we love hanging out together :)

What tasks await you

Developing ML Infrastructure A large team of strong ML specialists works alongside us. We need to provide them with a fast TTP (Time to Production), making it simple and quick to implement new models and experiment with them in the current infrastructure. For example: a framework for simulating feature state in production for offline model experimentation.

Product Development Think through and implement the product roadmap, develop user interaction scenarios, and integrate them into our system. For example: cross-service design and implementation of a new type of ad campaign, with some new optimization or budget constraint, and propagate it throughout the entire system.

Developing High-Load Runtime On one hand, develop the service for stream data processing and prediction calculation. On the other hand — improve the service that uses these predictions for ranking and selecting the best banner. For example: accelerating bid delivery through new technology.

Improving Infrastructure and Service Reliability Grow the maximum availability and reliability of our services.

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

We expect that you

• Develop in C++
• Have developed and supported high-load runtime services with high performance and reliability
• Can work effectively in a team and solve complex technical tasks, are proactive and driven for professional growth
• Love infrastructure services that directly impact the business

Will be a plus

• Know Python and SQL
• Are familiar with technologies and tools for processing and analyzing large volumes of data (e.g., Apache Kafka, Spark)
• Have interacted with ML infrastructure