Description

IT B2C is the largest ecosystem within Sber. We are more than 8,000 people across 18 cities in Russia. We are engaged in the development and advancement of retail solutions, making the Bank's services more accessible, secure, and convenient.

We are waiting for you!

We are a team of experts united by a shared passion for artificial intelligence and recommender systems (RecSys). Our main goal is to create a modern, scalable recommender platform capable of anticipating user expectations, offering personalized recommendations at every stage of their interaction with the Sber ecosystem. Our solutions cover a wide range of industries: from finance and e-commerce to entertainment and healthcare.

The development of our platform is centered around implementing new SOTA models. We monitor global trends, experiment with new approaches, implement them as part of the platform, and bring them to specific business applications.

Generative AI is currently setting a new vector for RecSys development. We see the future in a paradigm shift for building recommender systems: a transition from the classical architecture consisting of candidate selection and ranking stages to generative models. This opens the door to true end-to-end learning and deep synergy with LLMs.

We are looking for a Senior Data Scientist to join our R&D team to create the next-generation recommender system. We have set an ambitious goal to develop and bring to implementation a generative recommender model, trained end-to-end. If you are drawn to the idea of being a pioneer and want to stand at the origins of new technology, join us!

Responsibilities

• analyze new approaches, review scientific papers, generate and test hypotheses to improve the quality of generative recommender models
• implement hypotheses in Python, write efficient and scalable code for training and inference of models on PyTorch, conduct experiments on a GPU cluster
• design experiment setups, targets, reward functions, quality metrics, prepare data for experiments
• fine-tune LLMs (Fine-tuning and Alignment) for the specifics of the recommender domain.

Requirements

• mathematical background
• 3+ years of experience in Data Science
• good knowledge of Deep Learning, practical experience in training models
• confident command of Python and PyTorch, experience working with key ML/DL frameworks (Hugging Face ecosystem, Lightning)
• understanding of the current state of the industry in Generative AI.

Will be a plus:

• experience in the field of RecSys or NLP
• experience in distributed training of large models on a GPU cluster
• experience in fine-tuning and validating the quality of LLMs, experience with frameworks like TRL/verl frameworks
• ability to work with big data (tables, time series) on PySpark.

Technology stack:

Python, PyTorch, Transformers, MLFlow

Conditions

• hybrid/office work format (optional)
• annual bonus and annual review
• extended private health insurance from day one + dentistry and preferential insurance for family
• Sber Corporate University, internal educational platform, participation in IT conferences
• office on Kutuzovsky Prospekt with recreation areas and a gym
• 90 days of remote work from any region of the Russian Federation (not applicable for support roles)
• preferential mortgage at Sber, corporate pension program, SberPrime+ subscription, discounts from partners and services of the group of companies.