Ural Logistics is a reliable partner in the field of freight forwarding services for railway transport. We invite an AI Engineer to our team to participate in the creation of our own system for processing documents and data for freight railway transportation and related services.
What you will do:
- Design and develop document processing pipelines: loading, normalization, segmentation, structure determination, and context preparation
- Extract structured data from contracts, addenda, additional agreements, acts, and other related documents
- Work with long context, tables, overlapping wording, and time-dependent changes in conditions
- Ensure attribution of results to the document, page, and original fragment
- Build hybrid search: full-text and vector retrieval, filtering, re-ranking, and context aggregation
- Develop a graph representation of documents, entities, and relationships for search and analytical scenarios
- Design agent and workflow scenarios: state management, tool calls, retries, fallback mechanisms, and failure handling
- Create reference datasets, field metrics, and regression tests for evaluating extraction, search, and generation — so that changes in models, prompts, or retrieval strategies can be tracked for improvements and degradations
- Develop Python services, APIs, background tasks, and integrations with internal sources
- Participate in model selection, experiment design, fine-tuning, and local inference optimization
What is important to us:
- We consider candidates with confident industrial development in Python at the middle+ level and above
- Experience with FastAPI or similar frameworks, asynchronous execution, queues, and background processing
- PostgreSQL and SQL at a working level
- Practical experience in deploying LLM solutions to production
- Understanding of RAG, structured output, tool calling, and multi-step model scenarios
- AI system for railway logistics AI · ML · Full-stack
- Ability to measure quality: build reference datasets, select metrics, classify errors, and conduct regression testing
- Experience with Docker, Kubernetes, and CI/CD
- Readiness to complete a vertical task from specification and experiment to operation
Tech Stack:
- Python, FastAPI, Pydantic, PostgreSQL, Redis, Celery or equivalent, LangChain, LangGraph, MCP, RAG, GraphRAG, Qdrant, pgvector, OpenAI-compatible API, vLLM, Docker, Kubernetes
Cooperation Format:
- Full-time employment and the possibility of employment through a Russian legal entity as a staff member or through an agreed outstaffing model
- Main format — office or hybrid. For independent specialists, remote work from Russia is possible
- The final grade and scale of responsibility will be determined based on the interview results.
