Sber is looking for an AI Lead.
Sber AI's Directorate for AI Initiatives Implementation and Popularization, within the Department of Artificial Intelligence and Machine Learning Development, is seeking a DS/AI Lead to implement AI/GenAI projects in the B2G segment. The team assists the government in disrupting public administration through the adoption of AI/GenAI and accelerating national AI/GenAI development. It organizes one of the world's largest specialized events — AI Journey, unites business, government, and the international scientific community to drive the country's GenAI transformation.
Responsibilities:
- Leading prototyping in the NLP domain
- Developing and validating ML models, creating and implementing AI/GenAI solutions for B2G
- Designing AI/GenAI and ML model architectures for business tasks
- Formalizing business tasks into ML problems, selecting quality metrics and success criteria
- Monitoring deadlines and plans for pilot execution and prototype development
- Designing, developing, and quality controlling features, particularly multi-agent LLM-based systems for complex problem-solving
- Designing and developing RAG systems
- R&D in NLP: experimenting with the latest architectures, reasoning models, and fine-tuning models for specific tasks
- Optimizing inference for high-load operation
- Data collection, analysis, and preparation (EDA), feature engineering
- Conducting experiments, comparing approaches, interpreting results, preparing recommendations for the client
- Data analysis, feature engineering, A/B testing
- Participating in model deployment to production: collaborating with engineers, preparing artifacts (specifications, data/service requirements)
- Preparing technical documentation and presentations of results
- Interacting with business clients and technical teams
Requirements:
- Ability to solve ad-hoc tasks outside standard product development approaches
- Higher technical/mathematical/economic education
- 5+ years of experience in production ML system development
- Deep understanding of modern LLM architectures (Transformers, attention mechanisms)
- Practical experience with RAG: vector databases, embedding models, retrieval strategies
- Experience with distributed computing (PyTorch Distributed, DeepSpeed)
- Python proficiency for writing efficient production code
- Experience managing a development team, including those using automatic code generation systems (Codex, etc.)
- Understanding of the full ML model lifecycle, knowledge of key approaches to model evaluation, handling imbalance, overfitting, and interpretability
- English proficiency at Upper Intermediate level or higher
- Experience using GigaChat, Kandinsky, and similar tools in products, skills in creating and using AI agents
Will be a plus:
- Experience with information retrieval systems
- Experience building multi-agent systems (LangChain, AutoGen, CrewAI)
- Experience fine-tuning large language models
- Knowledge of modern approaches to prompt engineering and chain-of-thought reasoning
Conditions:
- Comfortable modern office near Kutuzovskaya metro station
- Hybrid work format
- Annual salary review and annual bonus
- Corporate gym and recreation areas
- Over 400 educational programs from SberUniversity for professional and career development
- Extended voluntary medical insurance, preferential insurance for family members
- Employee mortgage at a rate up to 1/3 lower than the current rate
- Prime subscription with the option for shared use among three close contacts
