Advisory AI Engineer

The Advisory AI Engineer at Lenovo’s AI Technology Center turns enterprise AI agent system reference designs and prototypes into production products. The role focuses on back-end engineering, with front-end development capabilities, and delivers AI systems across Lenovo’s hybrid-cloud, edge and end-device environments.

Responsibilities

• Build highly available, scalable distributed services for AI agent systems; develop APIs, SDKs and core business logic.
• Develop front-end interfaces for PC, mobile and cloud use, and integrate them with back-end services.
• Containerize and orchestrate AI products, deploying them across AWS or Azure public cloud, private cloud and edge devices. Address cross-region compatibility, network latency and resource scheduling.
• Improve back-end, database, caching and front-end performance to reduce latency and increase throughput, accounting for device and server hardware characteristics.
• Write deployment and operations documentation, support delivered products, monitor their operation, and iterate on features and performance.

Required qualifications

• Minimum 5 years of Python back-end development experience, including asynchronous programming and high-concurrency I/O. Experience building gateway services with FastAPI or Starlette and using SSE, WebSocket and NDJSON streaming.
• Proficiency with Pydantic, type annotations, strict type checking, pytest, asynchronous testing, packaging and dependency management.
• Production experience with Redis and PostgreSQL, including connection pools, transactions, consistency, indexing, timeouts, retries, idempotency, recovery and migrations.
• Practical Docker and Docker Compose experience; familiarity with Linux deployment and Shell scripting. Vue skills are mandatory, including independently building admin dashboards and debugging pages, integrating APIs and troubleshooting issues. Strong system design and cross-service diagnosis skills are required.

Preferred qualifications

• LangGraph or similar graph orchestration frameworks; state machines or workflow engines; OpenTelemetry observability.
• Production LLM application engineering involving multi-agent orchestration, tool calling and structured-output validation; vector search and memory integrations using tools such as pgvector, ONNX Runtime or tokenizers; and Langfuse or comparable evaluation platforms.
• Java, Python and C++ development proficiency, or experience maintaining open-source projects and their versions, compatibility, documentation and examples.