Kazakhtelecom JSC is a leading telecommunications company in Kazakhstan. Development of a corporate AI-agent platform, including RAG assistants for internal documents and business process support.
Responsibilities:
• Implement AI agents from concept to production; build LangGraph workflows and RAG systems; iterate prompts, connect tools and function-calling, and integrate external services; work with LLMs through LiteLLM and configure streaming; manage background tasks with Celery/Kafka; write pytest tests, assess response quality, and resolve production issues.
Requirements:
• Confident Python development with clean, maintainable code; practical LangGraph experience; LLM operations and application development; RAG with embeddings, vector search, relevance assessment, Qdrant, hybrid/dense/sparse search, and advanced chunking; Git and independent task completion; backend development with FastAPI/async, REST, PostgreSQL/SQLAlchemy, and queues; familiarity with LiteLLM/vLLM and local models such as Qwen or Gemma.
Optional:
• Experience with Celery or Kafka, Airflow, Docker/Kubernetes, CI/CD, S3/MinIO, and Keycloak; LLM quality assessment and observability with Langfuse or MLFlow; prompt engineering, multilingual RU/KZ/EN data, OCR/document processing, clean architecture, DI, DDD, and model fine-tuning.