Senior AI R&D Efficiency Architect

Responsibilities

• AI Engineering Operating System Construction : Define division-wide AI-consumable standards (R&D workflows, PRD templates, LLM I/O specifications, design tokens), transforming "context infrastructure" into an organizational compounding asset.
• Context & Memory System Development : Build a global Spec repository, Skill asset library, Golden Dataset, and unified Eval framework, ensuring every piece of knowledge accumulation compounds over time rather than starting from scratch each cycle.
• Multi-Agent Workflow Orchestration : Integrate external AI tools or self-developed Agents into a unified platform; design end-to-end Agent Pipelines; drive Agentification of high-frequency business processes and promote cross-team reuse.
• AI Risk Governance & Quality Gates : Establish Agent behavior boundaries, hallucination interception, and Fallback mechanisms; build automated quality gates for AI-generated code, ensuring every stage's output is verified before it flows downstream.
• Developer Enablement & Organizational Promotion : Encapsulate platform capabilities into out-of-the-box Agent services; drive the team's evolution toward an AI-native organization and build sustainable, compounding engineering moats.

Qualifications

• Deep hands-on experience in R&D efficiency, automated testing infrastructure, and CI/CD system development, with proven capability to build foundational platforms from 0 to 1.
• A clear understanding of the capability boundaries of large language models; hands-on project experience in Agent architecture design and deployment is a strong plus.
• Familiarity with cutting-edge AI coding tools such as Claude Code and Cursor; experience building complete AI systems combining "model + memory + workflow orchestration + evaluation loops" is preferred.
• Strong systems abstraction capability and a holistic perspective; ability to drive organization-wide standard adoption across teams without formal authority.
• Background at a top-tier internet company with architect or technical expert experience; awareness of model capability boundaries is a plus.