Data Science

We are looking for a Data Scientist Middle+ and above Registration as an Individual Entrepreneur Hourly rate 1200 rub/hour (200k/month) Full-time workload Remote Tasks: Designing architecture and developing LLM agents using LangChain, LangGraph, and internal Python SDK. Deploying and orchestrating agents in a production environment (Docker, Kubernetes). Developing quality evaluation pipelines (Evals): designing test scenarios, synthetic datasets, and metrics to check accuracy and absence of hallucinations. Integrating agents with internal corporate systems via REST/gRPC API and message brokers (Kafka/RabbitMQ). Optimizing model inference and managing GPU resources to reduce latency. Requirements: Python: Deep understanding of GIL, asynchronous model (asyncio, trio), descriptors, metaclasses, and garbage collector behavior. Concurrency: Practical experience managing threads and processes (threading, multiprocessing, concurrent.futures), implementing queues, semaphores, graceful shutdown, and protection against cascading failures. Backend: Developing high-load APIs (FastAPI, Pydantic v2), implementing complex authorization (OAuth2/JWT), introducing contract testing (TDD) and SLO/SLI monitoring. Resource Management: Direct interaction with system calls, configuring cgroups/cpuset, monitoring RSS/VMS, managing file descriptors (ulimit), and protecting services from OOM Killer. AI/LLM: Experience with fine-tuning and retraining models, working with embeddings, vector databases (Weaviate, ClickHouse), and evaluation frameworks (RAGAS, LlamaIndex). Tools: Linux (systemd, journald), Docker, Kubernetes (Helm), Prometheus/Grafana, GitLab CI. Contact: tg recruiter contact