RocketHuntSmart job search
Сбер
Russia3 months ago
Reach out directly about this role
Work Format
Onsite, Hybrid
Employment
Full-time
RocketHunt estimate from real salaries
AI writes it from your resume. You only have to send it.
The AI-Hub team of the Finance Block develops and maintains a platform for the development, storage, and execution of ML models and AI agents. We provide centralized APIs for interacting with models and development tools that meet Sber's architectural and cybersecurity requirements.
We are looking for an engineer who can analyze manual support processes and transform them into reliable software and AI tools.
Analysis of technical and business support processes, identification of manual operations, bottlenecks, and scenarios for automation
Design and implementation of support processes: from scripts and internal services to LLM and agent solutions
Testing, monitoring, and quality assessment of automated solutions; error analysis, development, and reliability improvement
Development and support of LLM components: system instructions, prompts, knowledge bases, execution scenarios, and skills
Building self-service tools for deploying ML models and AI agents in a production environment
Proficient skills in Linux, Git, HTTP/REST API, SQL, Docker, Kubernetes
Solid knowledge of Python (backend application development with FastAPI, ML model training)
Understanding of LLM operational principles and generative model limitations
Experience integrating LLMs into application systems
Understanding of RAG: indexing, document chunking, embeddings, search, reranking
Understanding of agent scenario building principles
Understanding of risks: hallucinations, prompt injection, data leakage, uncontrolled agent actions
Will be a plus:
Experience in developing CI/CD pipelines
Experience configuring and working with Helm charts
