Description
Wealth Management Transformation
About the project: Our team is implementing a strategic task to transform the business into an AI assistant for clients in finance and insurance. We are creating an ecosystem of services:
- Financial Advisor: Consultations, goal setting, expense analysis, planning, and saving.
- Personal Insurance Consultant: Assistance in product selection, risk assessment, conscious management of personal protection and capital protection.
We manage 28 Middle+ level and above experts (ML/DS/DA/DE).
We are looking for a strong AI solutions developer to expand the team.
Responsibilities
What you will be doing:
- AI Agents Development: Designing and implementing multi-agent systems (Prompt Engineering, RAG, Tools, Function Calling) into real banking processes.
- LLM Work: Optimizing and Fine-tuning models for the business domain (SFT, RLHF), working with open-source and proprietary models.
- MLOps & Production: Deploying models to PROD, building data processing pipelines on GPU clusters, organizing model quality and drift monitoring.
- Quality Assessment: Developing generation quality metrics (LLM Eval), conducting A/B tests.
- Integration: Developing services in Python (API), interacting with the development team and product analysts to control and provide expertise on data and functionality requirements.
Technology stack: Python, PyTorch, Hugging Face Transformers, LangChain / LangGraph, FastAPI, PySpark, Hive, Airflow, VectorDBs.
Requirements
- Experience: 2+ years of commercial development as a DS/ML Engineer (including at least 1 year of intensive work with NLP/LLM).
- LLM & NLP: Deep understanding of transformer architecture, hands-on experience with RAG, context windows, inference optimization methods.
- Development: Solid knowledge of Python (asyncio, aiohttp/FastAPI), experience writing clean, maintainable code.
- Big Data: Experience with distributed computing (Spark/PySpark, Hive) and orchestration (Airflow).
- Infrastructure: Understanding of MLOps processes, experience deploying services on GPU (on-premise or Cloud).
- Security: Understanding of personal data handling principles and security when using LLM in enterprise.
Conditions
- Full compliance with the Labor Code of the Russian Federation, preferential terms for insurance and loan programs for Bank employees ;)
- Schedule 5/2, office at Kutuzovskaya metro station;
- Hybrid work mode
- Competitive compensation package (salary + substantial annual bonus);
- VHI from the first day of work and preferential medical insurance for close relatives;
- Social support for employees;
- Opportunity to become part of a team implementing the bank's digital transformation programs - based on best practices and innovative ideas;
- Broad opportunities for professional development: corporate university and numerous online learning programs. Corporate training at company expense;
- Free SberPrime+ subscription, numerous discounts and bonuses from partners: SberMarket, MegaMarket, Samokat, Eapteka, etc.;
- Corporate pension program
