Data Scientist

Middle+/Senior Data Scientist (Python / AI agents / RAG / FastAPI / LangGraph) 🧠🤖

Vacancy ID: 91825_SS Rate: To be discussed. Grade: Middle+ / Senior. Workload: Full-time. Engagement Period: Until the end of 2026 with the possibility of extension. Format: Remote. Time Zone: UTC+3 (MSK). Location/Citizenship: Russian Federation. ⏳ Vacancy Validity Period: Until 20.10.2026. 📌 Important: Candidates with an experience relevance of 85% or more are considered (outstaffing position). 🏢 About the project: Development of internal products for an IT company with a focus on AI solutions. 🧩 About the role: This is a position for a Data Scientist with a strong engineering background. We are looking for a specialist who can not only build RAG and agent systems but also deeply understands Python (down to OS memory and resource management), can design fault-tolerant FastAPI services, and develop SDKs for agent solutions. 🚀 Key tasks: — Creation of AI agents using LangChain and LangGraph. — Design and development of an internal SDK to support agent solutions. — Formation and preparation of evaluation datasets (eval datasets) for testing and validation of agents. — Optimization of LLM inference and implementation of fault-tolerant microservices. 🎯 Mandatory Requirements (Hard Skills): Python Core (CRITICAL!): — Deep understanding of language fundamentals: working with data types, functions and classes, exception handling, iterators and generators, context managers, typing module. — Knowledge of the internal structure of objects and memory management principles. Parallel and Concurrent Processing: — Proficiency in async/await and asyncio. — Working with threads and processes, using queues and semaphores. — Implementation of task cancellation mechanisms, setting timeouts, and load control. Backend and Fault Tolerance (CRITICAL!): — Development in FastAPI, data validation via Pydantic. — Implementation of authentication mechanisms, application of TDD methodology. — Setup of logging, auditing, metric collection, and tracing systems. System Resource Management: — Working with subprocess and multiprocessing, signal handling. — Setting CPU and RAM limits, controlling RSS and file descriptors. — Handling frozen processes and preventing OOM. AI/ML and Agents (CRITICAL!): — Proven experience in implementing RAG solutions in corporate systems. — Understanding of agent system construction principles: ability to implement the agent loop in pure Python. — Knowledge of key AI agent architecture elements and their interaction. — Practical skills in LLM inference optimization (model and engine tuning to increase throughput). Infrastructure: — Experience deploying microservice architecture: Docker and Kubernetes. 💡 Will be a strong plus: — Experience with LangGraph for building complex multi-step agent loops. — Experience in creating custom frameworks/SDKs for AI tasks. — Practice in building eval pipelines (Ragas, TruLens, DeepEval, or custom metrics). — Experience in LLM inference optimization (vLLM, TensorRT-LLM, ONNX Runtime). — Knowledge of vector databases (Qdrant, Milvus, pgvector, Weaviate).

📲 Contact for application: recruiter contact (When applying, please immediately send your updated resume with an emphasis on deep Python knowledge, experience in building RAG and AI agents (LangChain/LangGraph), FastAPI, and LLM inference optimization. Please include the vacancy ID: 91825_SS)