Associate Director - AI Infrastructure Architect (AI Factory)

Start free →

Or apply yourself on the official page →

About the role

Job Description Summary Novartis architecture principles call for AI adoption guided by business value: AI is evaluated for new solutions and enhancements wherever it delivers a clear benefit, and it must be delivered on infrastructure that is secure, compliant and economically sustainable. Within Cloud & Hosting Services, AI & HPC Engineering is the single product management and delivery channel for AI infrastructure and AI use-case onboarding. The AI Infrastructure Architect sets the architecture those platforms are built to. The role defines the reference architecture and guardrails for AI workloads across Azure, AWS and private AI environments: training and inference platforms, GPU capacity and scheduling, data pipelines and data access, model operations, logging, monitoring, cost control and lifecycle governance. It establishes the standard intake pattern by which an AI use case moves from idea to a hosted, monitored, auditable service, so that scientists and business teams get speed without the organisation inheriting uncontrolled risk. This is an individual contributor architecture role. Its authority comes from design quality, evidence and adoption rather than from line management.   Job Description Key Responsibilities: • Own the AI infrastructure reference architecture and the standards that govern training, fine-tuning, inference and retrieval-augmented patterns across public cloud and private AI hosting. • Define the standard AI workload intake pattern, including architecture review criteria, data classification handling, environment selection and the conditions under which a use case may go to production. • Set guardrails for sensitive data, model access, prompt and inference logging, monitoring, resilience, retention and cost, in partnership with Security, Data Privacy and Quality. • Design GPU and accelerator capacity patterns - shared pools, reservations, scheduling, quotas and burst to cloud - and support the associated capacity and cost models. • Support qualification of AI platforms, regions and services for enterprise consumption together with AI & HPC Engineering and the public cloud architects. • Define the MLOps and LLMOps reference chain: model registry, versioning, evaluation, deployment, drift and performance monitoring, and decommissioning. • Maintain architecture decision records for AI platform choices, linking each to business value, risk posture and total cost of ownership. • Advise business, research and Infrastructure Solution Delivery stakeholders on feasible AI infrastructure options, and translate scientific and commercial requirements into technical designs. • Track the AI infrastructure market, evaluate emerging accelerators, model-serving stacks and platform services, and bring credible innovation into the roadmap. • Coach engineers and architects on AI infrastructure practice and act as a role model for the Novartis Values and Behaviours. Essential Requirement : • 8+ years' experience in infrastructure, cloud, or platform architecture, with a strong track record of designing and supporting enterprise-scale technology solutions. • Demonstrated expertise in building, deploying, and operating AI and machine learning infrastructure within production environments. • Experience enabling AI solutions to transition from proof-of-concept stages into scalable, governed, and business-critical services. • Strong knowledge of GPU and accelerator technologies, Kubernetes, container orchestration, and cloud-based AI platforms such as Azure and AWS. • Hands-on experience with model serving, inference optimization, MLOps/LLMOps practices, and modern data and vector pipeline architectures. • Solid understanding of identity and access management, observability, cost optimization (FinOps), an

View the official posting →