Description

We are building the core of a new AI platform for automatic assessment of companies' operational risks. Our goal is to replace labor-intensive manual processes by using coordinated systems of LLM agents that analyze multiple heterogeneous data sources, make informed decisions, and generate reports.

We need an experienced Python engineer who is capable not just of writing code, but of designing scalable solutions—from hypothesis testing to deploying production-ready agentic systems that solve specific business problems.

Why it's interesting to work with us:

• Real scale and impact: we are creating a new paradigm for managing non-financial risks autonomously, without people, instantly and reliably.
• Strong team and growth: design sessions, knowledge sharing; opportunity to grow into a Staff/Tech Lead role or deepen expertise in system/ML design.
• Modern processes: latest versions of Python and AI frameworks, short development cycles, CI/CD, and minimal bureaucracy.
• Full focus on development: a dedicated DevOps team handles infrastructure tasks.

Responsibilities

• Design and develop Python backend services within a multi-agent system (low latency, high reliability, observability).

• Integrate and develop the architecture of agentic systems (A2A, MCP, RAG, guardrails, evaluation).

• Work with data and integrations: queues/streaming, databases, cache, external APIs.

• Test hypotheses, quickly prototype, evaluate, and bring solutions to production.

• Write unit tests, set up monitoring and CI/CD.

• Mentor and conduct code reviews.

Our tech stack:

Python 3.12+, FastAPI, Dishka, Triton/vLLM, PyTorch, LangChain/LangGraph, gRPC/HTTP, Kafka, PostgreSQL/PgVector, Grafana/OpenTelemetry.

Requirements

• 3–5+ years of professional backend development in Python.

• Experience working with databases and message queues (Postgres, Kafka).

• Experience working with LangGraph/LangChain.

• Experience developing high-load microservices.

• Culture of writing unit tests (pytest).

• Experience with monitoring and logging tools.

• High level of teamwork culture and meticulous attitude towards commits.

• Willingness to take responsibility for product outcomes, not just code.

• Willingness to learn new technologies and approaches, critically evaluate them, and bring them to production.

Will be a plus:

• Strong mathematical background, engineering taste, and a habit of measuring everything with metrics.

• Interest in LLM or classical ML and experience in developing such solutions.

• Experience with LLM model inference.

• Participation in open-source projects or publications/presentations.

• Knowledge of the subject area in risk management.

Conditions

• Comfortable office located at Kutuzovsky Prospekt, 32.
• Work format - hybrid.
• Sber's unique learning system for professional and career development, seminars, trainings, conferences.
• Adaptation program and manager support at the start.
• Voluntary Health Insurance (VHI), accident insurance, social guarantees.
• Flexible mortgage discount equal to 1/3 of the Central Bank's key rate.
• Free SberPrime+ subscription, discounts on partner company products.
• Referral bonus for recommending friends to join the Sber team.
• Corporate pension program.
• Corporate events.