ML Engineer at Yandex Travel

Yandex Travel is one of the leading services in Russia for searching and booking hotels, flights, and train tickets. We are looking for someone who is ready to dive into the specifics of dynamic pricing for unique properties, where traditional pricing strategies stop working.

What tasks await you

Working on a classification model to predict booking probability You will need to create an ML model that, based on user signals, the hotel card, and session context, estimates the probability that a user will proceed to booking within N hours or days. The model will be a universal decision-making layer: instead of treating all users or offers the same, you can optimize the search results, marketing, and monetization based on the expected outcome.

Working on a regression model with a custom loss function to optimize price The task is to build a regression model that, for each hotel and stay dates, predicts the optimal adjustment to the partner's or supplier's base price to maximize expected profit while maintaining competitiveness and conversion. And the custom loss function will help optimize what matters to the business: margin, conversion, revenue, and compliance with constraints.

Developing personalized discount logic Together with the product team, you will search for where an incentive trigger will actually lead a user to book. Together with the marketing team, you will need to exclude audiences with a low booking probability from campaigns and configure retargeting: choose who to show reminders and discounts to, and who not to spend the budget on.

More about ML at Yandex — in the channel Yandex for ML

We expect that you

• Have worked with ML and DL
• Have experience with A/B testing, causal inference
• Have an excellent knowledge of linear algebra, probability theory, mathematical statistics, operations research
• Are proficient in Python, SQL

It will be a plus

• Have worked in retail, solved problems in the fields of pricing, recommender systems (RecSys), automated machine learning (AutoML)
• Have evaluated real estate or transportation
• Have experience with Offline Reinforcement Learning
• Have a T-shaped profile