Intelligent Agent & Model Development Manager

In this position, you will:

• Lead the core technical architecture design of large model platforms and AI applications, deliver key project milestones, and integrate with external AI ecosystems (vendors / open-source communities). Build high-performance workflow engines and model scheduling systems to ensure the platform supports high concurrency, high availability, multi-model compatibility and fast delivery.
• Design the architecture for large model middle platform / Agent platform (including Gateway, Routing, RAG, Cache modules), and lead capability building and technology selection for PaaS/SaaS-grade AI platforms.
• Oversee major technical projects related to enterprise large models, formulate agile development plans, conduct core technology research, and control full-lifecycle schedule and code quality.
• Drive R&D of low-code / no-code AI workflow engines (Workflow Engine), enabling flexible orchestration across Prompt, Agent, Tools, APIs and various business systems.
• Manage integrations with external large model vendors (API connection, private cloud deployment), track latest developments in open-source communities such as LangChain, LlamaIndex and vLLM, and introduce cutting-edge technologies to reduce costs and boost efficiency.

To be successful in this position, you should meet the following requirements:

• 5 years or above relevant working experience
• Bachelor’s degree or above in Computer Science, Communication Engineering, IT, Software Engineering, Information Security, Big Data, Artificial Intelligence or related disciplines.
• Expertise in high-concurrency and microservice architecture; hands-on experience with Kubernetes ecosystem and GPU computing resource scheduling is preferred.
• Familiar with at least one workflow / rule engine; proficient in LLM API calling mechanisms, streaming transmission (SSE) and Function Calling.
• Hands-on experience deploying and accelerating inference for mainstream open-source LLMs (Kimi, Deepseek, Llama, Qwen). Experience with vLLM, TensorRT-LLM is a strong plus.
• Strong governance awareness, pursuit of elegant code and high system availability, with excellent cross-team collaboration capabilities.