AI Systems Architect

About the role

Kyndryl is hiring an AI Systems Architect in Madrid. The role designs and evolves an enterprise-scale Agentic AI platform so that engineering teams can build, deploy and operate production-ready AI agents in a secure, governed and reusable way. It is a senior platform architecture role, not an application development role.

Key responsibilities

• Architect a reusable enterprise Agentic AI platform, including agent runtime, orchestration, context management, memory, evaluation, guardrails and observability.
• Develop and maintain an Agent Harness with standardized runtime capabilities and Model Context Protocol (MCP) integration.
• Define "Golden Paths" that standardize AI agent development and deployment across engineering teams.
• Build a model-agnostic AI runtime supporting multiple LLM providers, AI gateways and enterprise integrations.
• Collaborate with infrastructure teams on Kubernetes, GPU platforms, networking and cloud-native AI infrastructure.
• Establish enterprise AI governance covering policy enforcement, security, prompt and model lifecycle management, human-in-the-loop workflows, evaluation and cost optimization.
• Integrate AI agents with enterprise platforms such as CI/CD, ITSM, observability, cloud services and business applications.
• Define engineering standards and mentor teams to drive enterprise adoption of Agentic AI best practices.

Required profile

• 5+ years designing cloud-native platforms or enterprise software architectures.
• Strong hands-on experience with Agentic AI architectures, orchestration, tool integration, memory, evaluation and guardrails.
• Experience with agent frameworks such as LangGraph, CrewAI, Semantic Kernel or AutoGen.
• Expertise in Kubernetes, containers, Infrastructure as Code and cloud-native architectures.
• Understanding of AI infrastructure, including GPUs, model serving, AI gateways and observability.
• Experience designing developer platforms and reusable services rather than standalone applications.
• Professional proficiency in both English and Spanish.

Required skills

• Strong Python software engineering skills.
• Preferred: LiteLLM, OpenRouter, NVIDIA NIM, vLLM or Triton Inference Server.
• Preferred: LLMOps, AI governance, RAG, vector databases, semantic search, Prometheus, Grafana, OpenTelemetry, Langfuse, and AI coding assistants.

The role sits within the Kyndryl Consultant Profession, with a hybrid-friendly culture and access to learning, training and career development.