Staff Software Engineer, AI Platform

What you'll do

• Build and maintain an internal AI platform that provides a unified gateway for multiple model providers.
• Design the model gateway to handle provider abstraction, routing, failover, and seamless configuration changes.
• Develop evaluation, benchmarking, and guardrail capabilities to ensure safe deployment of agentic systems.
• Manage inference costs and latency as primary engineering concerns including spend attribution and routing logic.
• Establish technical standards for API design and infrastructure that other internal teams will build upon.
• Ensure the platform meets corporate requirements for security, compliance, and responsible AI standards.
• Solve complex integration problems and provide high-level technical guidance to other engineering teams.
• Mentor distributed team members by creating durable designs and documentation for the platform's evolution.

What we're looking for

• Must have 8+ years of experience building and operating production systems.
• Must have direct experience shipping agentic systems to production including retrieval, tool calling, and cost/latency management.
• Must have hands-on experience with more than one model provider.
• Must have experience with evaluation, observability, or guardrail tooling for agentic systems.
• Must have a strong platform engineering track record on systems where other teams were the users.
• Must have hands-on depth in at least one major public cloud for running and debugging production workloads.
• Must have experience designing stable APIs and abstractions that other engineers build against.
• Must demonstrate production ownership including on-call, incident response, and root cause analysis.
• Must possess excellent English communication skills, both written and spoken.
• Experience with internal developer platforms or multi-tenant services (preferred).
• Experience with Kubernetes, container-based delivery, and infrastructure as code (preferred).
• Experience with enterprise security, data privacy, or responsible-AI review processes (preferred).