AI Agentic Solution Architect

About the role

Roche is looking for an AI Agentic Solution Architect in Shanghai to drive technological innovation and deliver strategic business value for Roche China. You will lead the end-to-end design, development, iterative enhancement and ecosystem integration of the AI infrastructure and platform, keeping all systems strictly aligned with enterprise security and compliance standards. The role blends enterprise system design, platform engineering and agile delivery with deep AI expertise.

Key responsibilities

• Design and manage agent identity and registration systems, scalable agent runtimes and internal/external agent marketplaces.
• Build agent gateways for routing and guardrails, and implement observability, evaluation frameworks and optimization loops.
• Integrate tools, APIs and Model Context Protocol (MCP) systems, and design vector and graph database contexts for RAG architectures.
• Architect AI and agentic applications using LangChain and LangGraph, and enforce domain-driven design, REST/gRPC/GraphQL, Kafka/RabbitMQ and caching.
• Define non-functional requirements, SLOs, shift-left testing, security guardrails (prompt injection defense, RBAC for RAG) and FinOps strategies.
• Manage Kubernetes clusters, infrastructure-as-code with Terraform and CloudFormation, and observability (metrics, logs, tracing).
• Lead large-scale projects with PMM and Agile/Scrum, coordinate cross-functional teams, and author architectural documentation.

Required profile

• 10+ years as a Solution Architect, with at least 2+ years of hands-on AI architecture or AI platform design.
• Experience in pharmaceutical, life sciences or top-tier tech is highly preferred.
• Fluent English and Mandarin.

Required skills

• Agentic AI with LangChain and LangGraph, and RAG architecture.
• Cloud-native architectures (AWS, Alibaba Cloud) and Terraform, Ansible, Packer, CloudFormation.
• CI/CD pipelines with GitLab and JFrog Artifactory.
• Kubernetes, HA, disaster recovery, RBAC, SSO and LDAP.