Description
We are building an AI platform from scratch that automates complex analytical tasks using LLM and agent architecture.
The system independently searches and combines data from various sources, identifies hidden connections, and forms substantiated conclusions. It is based on a multi-agent approach, RAG, memory, a relationship graph, and proprietary data processing pipelines.
This is not an integration of a "wrapper over an LLM," but the development of a full-fledged intelligent backend core, where you can influence the architecture, experiment, and create a complex deep-tech-level system.
Responsibilities
Tasks:
- design and implement an AI agent based on LLM (Python, LangChain/LangGraph), including support for RAG, reasoning chains, and interaction with external systems
- implement agent capabilities in a reactive loop: memory (short- and long-term), planning
- develop REST (FastAPI/Flask) for integrating the agent with internal services and DBMS
- conduct prompt engineering and create validation scenarios, assessing the quality and safety of the agent's responses
- integrate the agent into the corporate infrastructure: Jenkins, OpenShift/Kubernetes, CI/CD pipelines.
Requirements
We expect from the candidate:
- higher education in an IT field
- understanding of the principles of LLM, embeddings, RAG, and modern approaches to agent architecture
- confident proficiency in Python and experience in backend service development - from 2 years
- experience with LangChain, LangGraph, FastAPI/Flask, and familiarity with Hugging Face
- practical experience in creating AI agents with access to external data — experience with multi-agent systems will be a particular plus
- knowledge of REST API principles, ability to design reliable and scalable APIs
- basic skills in working with PostgreSQL and understanding of SQL
- experience using Git, Jenkins, Docker, Kubernetes/OpenShift.
Will be a plus:
- experience working with big data storage systems: ClickHouse, Hadoop
- basic proficiency in PyTorch
- experience with graph databases.
Conditions
We offer:
- a comfortable modern office near Kutuzovskaya metro station
- the opportunity to choose a convenient schedule – office/hybrid
- annual salary review, annual bonus
- corporate gym and relaxation areas
- over 400 educational programs from SberUniversity for professional and career development
- extended DMS, preferential insurance for family, and a corporate pension program
- free SberPrime+ subscription, discounts on products from partner companies
- referral bonus for recommending friends to the Sber team.
