AI Architect - RDT Artificial Intelligence — Roche

Full description

At Roche you can show up as yourself, embraced for the unique qualities you bring. Our culture encourages personal expression, open dialogue, and genuine connections, where you are valued, accepted and respected for who you are, allowing you to thrive both personally and professionally. This is how we aim to prevent, stop and cure diseases and ensure everyone has access to healthcare today and for generations to come. Join Roche, where every voice matters.

The Position

AI Architect - RDT Artificial Intelligence

This is a rare opportunity to shape the architectural foundation of enterprise AI at one of the world’s leading healthcare and diagnostics companies. As AI Architect within RDT Artificial Intelligence, you will hold strategic and engineering ownership of Roche’s enterprise AI technology stack - defining how AI is built, governed, and scaled across the organization.

This role serves as the central technical visionary connecting advanced agentic workflows, traditional AI and ML platforms, and core infrastructure. You will architect a secure, robust, and highly scalable enterprise platform ecosystem that enables diverse teams across Roche to deploy cutting-edge AI capabilities safely, compliantly, and cost-effectively - at global scale.

More than a platform role, this is a mission to prevent technological fragmentation and agentic chaos: to establish the digital guardrails and architectural principles that allow AI to proliferate responsibly across one of the world’s most complex regulated enterprises. You will translate strategic vision into concrete blueprints, and blueprints into production reality.

The Opportunity

1. Enterprise AI Architecture & Taxonomy

• Capability-Based Reference Architecture: Architect and evolve the enterprise AI reference architecture based on logical capabilities and flows, maintaining vendor neutrality and enabling multi-platform configurations rather than forcing monolithic alignment.
• Platform & Application Taxonomy: Govern the enterprise AI landscape, clearly demarcating developer-facing platforms (e.g., managed ML environments, API gateways, enterprise retrieval services) from ready-to-use business applications - ensuring every team can navigate the landscape without duplication or confusion.
• Advanced Technology Integration: Lead the technical design, proof of concepts, and production implementation of core developer enablement modules, including the AI Gateway, PI/PII detection pipelines, and advanced content safety layers.

1. Infrastructure Engineering & Compute Operationalisation

• High-Compute Software Framework: Take technical ownership of designing, validating, and establishing the production-ready software layer - orchestration, monitoring, job scheduling, and cost tracking - required to operationalise high-compute environments such as dedicated GPU infrastructure (e.g., NVIDIA AI Factory).
• Hyperscaler Strategy: Architect cloud-native multi-region endpoints across AWS, Azure, and GCP, ensuring optimal workload placement for model serving and training.
• Vector & Data Layer Strategy: Define structural standards for data and context layer integration, utilising specialised databases (e.g., Qdrant Cloud for long-term memory, PostgreSQL pgvector, MongoDB Atlas) linked natively into agent frameworks.
• Agentic Strategy: Translate the vision of an AI-augmented workforce into a structured agentic architecture - balancing speed of delivery with the stability, compliance, and safety controls required in a regulated enterprise.

1. Governance, Traceability & Security

• End-to-End Traceability Framework: Enforce enterprise invocation standards using structured Correlation IDs (business journey mapping), Trace IDs (independent agent execution), and Span IDs (sub-flows and tool calls) across the AI Gateway - making every AI interaction auditable and debuggable.
• Agent Governance Architecture: Author and execute technical pre-deployment sign-offs, agent identity propagation