AI Platform Engineer

This role is for a Senior AI Platform Engineer at Plarium, part of the Midcore District. The engineer will own Playamp's AI platform infrastructure, working with DevOps, Security, Engineering, and Product teams to build and scale AI-powered solutions. Responsibilities include designing and operating the internal AI platform, productionizing AI infrastructure on GCP, and bringing AI to DevOps workflows.

• Design, build, and operate Playamp's internal AI platform - model gateway, agent orchestration, RAG pipelines, vector stores, and the MCP servers that connect LLMs to our internal systems.
• Productionize AI infrastructure on GCP (Vertex AI, GKE, managed and self-hosted inference) using Terraform and GitOps.
• Bring AI to our DevOps and automation workflows.
• Own the agent lifecycle in production: registry, versioning, observability (tracing, evals, cost tracking), and regression gates.
• Carry standard senior DevOps responsibilities alongside the team: production ownership, on-call, networking, security hardening, and incident response on AI platform's core infrastructure.
• Develop guardrails that help the security teams track and monitor AI usage across the company.

• 5-7 years of infrastructure, DevOps, or platform engineering in production, including 2+ years dedicated to AI infrastructure (real systems, not POCs).
• Practical experience with the Model Context Protocol (MCP) and RAG - building or integrating MCP servers and exposing internal systems to LLMs.
• Experience designing and shipping agentic systems in production: multi-step, tool-using agents with guardrails, retries, and evaluation.
• Deep cloud experience, preferably GCP, with solid Kubernetes, networking, Infrastructure as Code (Terraform), CI/CD, and GitOps fundamentals.
• AI evaluation infrastructure: Built or owned eval harnesses for LLMs/agents - golden datasets, offline + online evals, regression gates in CI, A/B testing of prompts and agents in production.
• Strong Python. Hands-on with the modern LLM serving stack and at least one industry-standard agent framework (e.g., Google ADK, LangChain, or similar).
• Cost engineering for AI: practical experience managing AI spend in production - prompt and semantic caching, model selection trade-offs, batch vs real-time routing, per-team budgets and showback.