Senior AI Infrastructure Engineer

IronGrid Data Services is seeking a Senior AI Infrastructure Engineer to serve as a technical cornerstone of their AI Infra team. The role involves driving the design and execution of workflow systems that bridge platform APIs and AI-powered customer experiences, with a focus on building and owning production-critical systems.

Responsibilities:

• Own multi-tenant tool layers. Design and implement multi-tenant isolation, rate limiting, and caller-attribution systems as direct enterprise customers use MCP tools
• Extend A2A interfaces and experiences. Enable customer and internal agents to invoke other agents in support of new initiatives
• Build and own durable agent runtimes, including long-running execution, task lifecycle management, and failure recovery
• Lead evaluation-driven development. Design golden scenario suites, automated CI/CD evaluation pipelines, and regression detection across all workflows. Evaluation is a parallel engineering workstream, not an afterthought
• Implement progressive discovery strategies for context, tools, agents, and skills, using deferred loading, search, categorization, and semantic filtering
• Design the agent harness and orchestration loop, leveraging isolated context, shared memory, and multi-agent coordination
• Drive new agentic capabilities from design to launch for customer experiences, owning production rollouts
• Participate in on-call rotation for customer-facing production systems; contribute to incident response, postmortems, and reliability improvements. This team practices full-lifecycle ownership
• Partner with internal engineering teams to negotiate and promote API-first designs that serve both programmatic and agentic consumers
• Contribute to multi-provider LLM abstraction layers, ensuring flexibility among LLM providers
• Facilitate AI office hours and cross-team enablement, serving as the internal expert on LLM tooling, agentic patterns, and tool configuration for teams across VideoAmp engineering
• Author, review, and drive clear technical requirements documentation for new solutions: design docs, architecture diagrams, and postmortems