We are looking for a Data Science specialist to implement modern ML and AI solutions in various processes across all business lines of Sberbank Insurance. We are seeking a professional who will develop the generative AI domain and create innovative solutions based on large language models. If you have ideas on how to make the lives of many Sberbank Insurance clients better – then we are on the same path!
Responsibilities
- Conducting exploratory data analysis (EDA) to identify key metrics and patterns, building models. Data-driven approach to development
- Development and implementation of AI agents based on LLM (SOTA open-source solutions), including participation in the creation and development of RAG systems (Retrieval-Augmented Generation)
- Designing application architecture and user request processing pipelines
- Working with LLM: writing and optimizing prompts (prompt engineering), managing interaction with the model
- Integrating solutions with corporate systems, ensuring their interaction
- Creating autonomous business processes within the company
Requirements
- Experience with Python and libraries: pytorch, pandas, numpy, SQLAlchemy, Scikit-learn, Matplotlib/Seaborn
- Writing maintainable, readable code; participating in reviews, discussions, and technical approach formation
- Practical experience in NLP: text vectorization, request classification, building RAG systems, fine-tuning language models, embeddings, etc.
- Or/and practical experience in CV: fine-tuning SOTA models for classification, detection, segmentation, identification, etc.
- Understanding of classical ML algorithms
- Understanding of neural network model training methods (transfer learning, fine tuning)
- Git skills, experience integrating solutions into industrial environments
- Instrumental proficiency in AI for analysis, generation, and automation
Will be a plus:
- LangChain/LangGraph
- Basic skills with OracleDB/PostgreSQL
- Understanding of PEFT methods for LLM/VLM fine-tuning
- Experience developing chatbots and conversational systems, knowledge of ML/AI solution production implementation stages
- Containerization skills (Docker) and experience with async/threading.
- Skill in working with generative AI models and experience in creating AI agents and using them in work
Conditions
- Hybrid work format, in a modern office in Moscow on Prospekt Mira (Olimpiyskiy 14, BC Diamond Hall)
- Favorable preferential mortgage lending conditions
- Free SberPrime+ subscription, discounts on products from partner companies: Okko, Sber Market, Sber Eapteka, and others
- Voluntary medical insurance from day one and preferential insurance for loved ones
- Corporate pension program
- Children's recreation and gifts at the Company's expense
- Company-sponsored training: online courses in Sber's Virtual School and unlimited library access, training at the Corporate University, training sessions, meetups, and the opportunity to obtain a new qualification
