AI Platform Engineer

Great American Insurance Group•Cincinnati, OH

22h•Hybrid

About The Position

The AI Platform Engineer is a hands-on engineering role building and operating the enterprise AI platform that powers our Software Developer, Data Scientist, and Knowledge Worker communities. You will design, implement, and run the platform services that make LLMs, RAG systems, and MCP servers safely consumable across the enterprise. You will work inside a small, high-leverage platform team where code quality, automation, and operational excellence are the expectation, not the goal. You will report to the AI Platform Delivery Leader.

Requirements

5+ years of software engineering experience with a strong development background. You write production code, not just scripts.
Hands-on experience with AI Foundry (or equivalent enterprise AI platforms) including model deployment, evaluation, and lifecycle management.
Practical experience building RAG systems end-to-end: embeddings, vector databases (e.g., pgvector, Pinecone, Weaviate, Azure AI Search), retrieval tuning, and evaluation.
Working knowledge of MCP server implementation and the patterns for exposing tools and data to LLMs safely.
Strong DevOps practice: Git-based workflows, CI/CD pipelines, containers, Kubernetes, IaC (Terraform or equivalent), and observability tooling.
Familiarity with LLM APIs, prompt engineering, token economics, and the failure modes of production LLM systems.
Security and governance instincts: you assume every endpoint will be attacked and every model call will be audited.
Platform engineering mindset: you build self-service capabilities with golden paths, not bespoke solutions for each consumer.

Nice To Haves

Familiarity with the P&C Insurance industry

Responsibilities

Build and maintain core platform services including the LLM gateway, model registry, RAG pipelines, vectorization services, and MCP server infrastructure.
Implement AI Foundry-based capabilities and integrate them with existing enterprise platforms (API gateway, IAM, observability, ECM).
Develop reusable Python and/or Java libraries, SDKs, and golden-path templates that let consumers onboard in under an hour.
Own the automation that provisions access, onboards models, runs compliance checks, and emits telemetry, so platform scale is never gated by team size.
Design and operate retrieval pipelines including chunking, embedding, vector store management, re-ranking, and evaluation.
Tune retrieval quality against real consumer workloads and publish evaluation harnesses that let consumers measure their own RAG implementations.
Integrate vectorization services with the ECM system and other enterprise data sources under the governance controls defined by the platform.
Build, host, and operate MCP servers that expose enterprise capabilities safely to LLM consumers.
Implement the authentication, authorization, rate limiting, and audit patterns that make MCP usage enterprise-ready.
Partner with consuming teams to define and publish the MCP contracts the platform supports.
Treat the platform as a product: CI/CD pipelines, IaC, automated testing, progressive delivery, and full observability are table stakes.
Own the on-call rotation for the services you build.
Write the runbooks, automate the toil, and drive incident postmortems to real fixes.
Build safety, governance, and compliance controls into the pipeline so they are enforced by default, not added later.