Description
IT B2C is the largest ecosystem within Sber. We are more than 8000 people in 18 cities across Russia. We are engaged in the development and enhancement of retail solutions, making the Bank's services more accessible, secure, and convenient.
We are waiting for you!
We are the SberAds Team - we create predictive models and high-load services for their application to display online advertising. All models are used in real-time ad display auctions:
- starting with identifying the user
- then evaluating whether to participate in an ad display request
- then determining the user's interests
- further selecting relevant ads for display
- and finally, determining which ad and with which bid we go to participate in the auction.
The team directly influences the quality of displayed ads, advertiser satisfaction, and the efficiency of SberAds.
We are looking for a Data Scientist
Our stack: Python, Go, S3, Spark, Hive, Airflow, MLFlow, Kafka, ClickHouse.
Responsibilities
- build/improve models for different parts of the entire pipeline
- handle the full cycle of DS/ML tasks
- participate in the creation and development of the advertising platform
- improve the model building process (from hypothesis formulation to model performance monitoring).
Task examples:
- build embeddings based on advertising descriptions. Add this data to conversion prediction models
- train a model for predicting bid distributions in auctions. Integrate the model with the current bid determination algorithm
- conduct experiments with new offline/online features in models. Deliver features to the service on Go (colleagues can help here if needed)
- implement a model that filters suspicious incoming traffic.
Requirements
- knowledge of classical ML and Deep Learning approaches
- experience in industrial use of RecSys methods
- experience in model development from problem to production support
- confident experience in development with Python and Pyspark, experience in writing production code
- knowledge of principles of distributed data processing
- ability to make data-driven decisions and substantiate one's position
- ability to find solutions to problems, decompose, explain, and monitor execution.
Will be a plus:
- specialization in ranking, search, or RTB tasks
- experience in development with Go
- successful kaggle experience.
Conditions
- office at Mayakovskaya metro station (Oruzheiny Lane)
- hybrid work format
- competitive compensation (salary and performance-based bonuses)
- VHI from the first working day, accident and severe disease insurance
- financial assistance and social support, corporate pension program
- discount programs from partner companies
- free fitness on the employer's premises
- external and internal professional training: seminars, trainings, conferences, corporate library.
