MCP Platform Architect

About Noah Holdings

Noah Holdings is a leading wealth management platform serving global Chinese high-net-worth clients with integrated advisory, asset allocation, and cross-border financial solutions. As the company continues to invest in digital transformation and global expansion, building an enterprise-grade AI capability platform has become a critical foundation for innovation, productivity, and scalable intelligent operations.

Key Responsibilities

• Lead the overall architecture and implementation of the MCP platform, adapting open-source MCP ecosystems into an enterprise-ready Server / Client framework for Noah’s business scenarios.
• Design and build the full WeCom-to-MCP-to-Claude-to-Skill execution flow, including message triggering, user identity mapping, authorization, async orchestration, timeout handling, retry policies, and response delivery.
• Establish a standardized Skill runtime environment covering registration, invocation, authentication, sandbox isolation, logging, tracing, auditability, and developer-facing operational tooling.
• Build governance mechanisms across the full platform lifecycle, including review workflows, version control, staged rollout, rollback strategy, change management, and deprecation processes.
• Continuously evaluate and onboard high-value MCP servers and ecosystem components, such as search, knowledge collaboration, code repository, and workplace productivity integrations.
• Produce clear onboarding documentation, APIs, conventions, and toolchains that reduce integration cost for internal teams and improve developer productivity.
• Partner closely with Skill engineers, product stakeholders, and security teams to optimize platform APIs, permission models, and developer experience.

Requirements

Must-have

• 5+ years of software engineering experience, with proven delivery in enterprise AI platforms, middleware, developer platforms, or infrastructure products.
• Strong understanding of MCP concepts and open-source implementations, including the protocol design around Tools, Resources, and Prompts, differences across major SDKs, and common real-world pitfalls.
• Hands-on experience integrating enterprise IM platforms such as WeCom or DingTalk beyond simple webhooks, with working knowledge of OAuth, application permissions, message routing, and organizational identity mapping.
• Good understanding of LLM platform integration patterns, including Claude Enterprise capabilities, operational constraints, and enterprise security considerations.
• Solid distributed systems and platform engineering fundamentals, with the ability to design for authentication, isolation, audit, resiliency, scalability, and reliability.

Nice-to-have

• Experience with AI agents, tool-calling frameworks, plugin ecosystems, or internal developer platforms.
• Experience building permission systems, sandbox execution environments, observability pipelines, or SRE practices for platform products.
• Active interest in the MCP open-source ecosystem and broader enterprise AI platform practices.
• A product-engineer mindset with strong empathy for internal developers and a focus on practical adoption, not platform building for its own sake.

What We Offer

• A chance to shape the core AI infrastructure of a leading enterprise during a decisive build-out phase.
• Direct ownership of platform capabilities that will define how Skills are developed, governed, and scaled internally.
• Meaningful, high-complexity engineering challenges with visible business impact and cross-functional influence.
• A pragmatic, high-accountability environment that values speed, ownership, and long-term platform thinking.