Analytics Developer for Yandex Maps Directory

Yandex Maps is one of the most well-known mapping services in Russia, already operating in dozens of countries.

In addition to building addresses, Maps provides information about organizations, and the Directory service builds and maintains a database of them. We help find organizations by category and location or name, learn about phone numbers, websites, hours of operation, and other information. Users can read and leave reviews and ratings, upload photos and videos about organizations, and their owners can manage information through a special Business account. Our goal is to maintain and develop a high-quality global database of organizations.

The Directory is an extremely complex system, which is analyzed by a dedicated group. Right now, we are creating and validating Directory quality metrics—answering the questions:

• Which organizations are we unaware of and why?
• How accurately do we know the location, phone number, and working hours of organizations?
• How can we collect information about organizations better and more cheaply?
• How do errors in the Directory database affect users?
• How can we tell from user behavior that we gave a good response to their query?
• In which cities and countries are competitors better than us, and specifically in what ways?
• How do we understand an organization's importance to a user?
• How do we combat fraudsters who create fake organizations or provide us with incorrect information?

Our call centers and surveyors collect and verify information in hundreds of cities, we operate on terabytes of data about user queries and actions, their routes, and visits to organizations. Based on this data, our analysts build a picture of the real world, give optimal recommendations to the business, improve and reduce the cost of our processes, and help the service grow into new countries.

Our stack includes:

• Technology stack: SQL, Python, NumPy, pandas, DataLens, CatBoost
• Training ML models based on regression or trees
• A/B experiments on real users
• Regular data delivery and monitoring processes

What tasks await you

Content completeness and accuracy metrics How well do we know the location, phone numbers, and working hours of organizations? How to reconcile contradictory information about an organization coming from different sources: surveyors, assessors, competitors, users? How to improve the quality of information collection—through training and verification of surveyors and assessors or by other means? You will be tasked with creating basic data quality metrics for transport: stops, routes, schedules.

Efficiency How to reduce the cost of obtaining information? Which information sources are most useful? How to optimally allocate the budget for information refinement? In which segments (cities and categories) will investments in information collection be most effective? How well do we handle user complaints and how can we do it better?

Online user behavior metrics Which data errors in Maps are most critical for users? How can we tell from user behavior that we gave a good or bad response to their query? How to build a model and predict how data collection in a given segment will affect user behavior?

We expect you to

• Have a good understanding of basic mathematics, probability theory, and statistics
• Have used Python and pandas for data analysis
• Think analytically, be able to formalize and decompose tasks, build hypotheses, establish relationships, and draw reasoned conclusions

Nice to have

• Good knowledge of SQL
• Understanding of ML basics
• Experience as an analyst or developer