DS Engineer for the Search Recall Team

About the Team

The team is responsible for retrieving listings at the L1 search stage — that is, for forming the initial candidate pool for the search results. We develop and improve systems that efficiently select relevant products from among tens of millions of items. Our work combines both simple heuristics and deep learning models. The team closely collaborates with other search divisions to ensure high quality, speed, and scalability. Our goal is to make the search results as accurate and useful as possible for the user from the first milliseconds.

Examples of Future Tasks:

• improving the quality of vector search and experimenting with new embedding architectures;
• developing LLM models for listing summarization and data enrichment;
• creating new candidate generators considering the specifics of individual categories and user scenarios;
• optimizing and developing existing candidate generators to improve search accuracy and coverage;
• participating in the design and development of an internal platform for accelerated creation and testing of candidate generators.

You will be responsible for:

• preparing data, conducting analysis, and formulating hypotheses to improve search quality;
• developing MLP models — from quick prototypes to production-ready solutions;
• working with backend services in Python and Go, participating in the design and development of new components;
• demonstrating product thinking and assessing the impact of changes on key metrics and user experience;
• launching A/B tests, monitoring the correctness of experiments, and analyzing results.

We expect you to:

• have extensive experience in Python development and understand the basics of parallel programming;
• possess confident skills in working with PyTorch and have successful experience in creating non-standard training pipelines;
• know the main DL architectures in your domain and understand the transformer architecture.

It would be great if you:

• write in Go or any other strictly typed language;
• have experience in distributed training;
• have experience in search and recommendation systems;
• have experience in deploying DL models to production and optimizing inference.

Working with us means:

• the opportunity to improve the experience of millions of users;
• interesting and challenging tasks at a large scale;
• a strong team that is always ready to help;
• the opportunity to learn and try new things, with powerful hardware for this;
• a training budget that can be spent on courses or professional literature;
• care for your health: from day one, you will have VHI with dentistry; a therapist and a massage therapist are available at the office;
• the possibility to work remotely or from offices in four cities in Russia.