> Markdown version of [/jobs/ext/4816864-ai-platform-engineer](https://www.wearedevelopers.com/jobs/ext/4816864-ai-platform-engineer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # AI Platform Engineer - **Company:** Cornerstone Barricades - **Location:** Cary, NC, United States (Remote available) - **Experience:** Expert - **Salary:** $128,960 - $131,040 - **Contract:** Permanent contract - **Skills:** LangGraph Framework, Java (Programming Language), Artificial Intelligence, Continuous Integration, Elasticsearch, Python (Programming Language), Reliability Engineering, OpenAI, Search Technologies, Pinecone, Model Garden (Google Cloud), QLoRA, LangChain, Retrieval-Augmented Generation, Large Language Models, Qwen, NeMo Guardrails, Llamaindex, Langfuse — LLM Observability and Analytics Platform, Rate Limiting, Pgvector, AI Platforms, LangSmith, Kubernetes, CrewAI, RAGAS (Retrieval Augmented Generation Assessment), Weaviate, AutoGen, Arize Phoenix, Ollama, LiteLLM, Machine Learning Operations, DeepEval, Model Context Protocol, Terraform, Promptfoo - **Published:** November 21, 2026 - **Apply:** https://zipstaff.betterteam.com/ai-platform-engineer/apply/standard ## About the Role * US citizen or green card holder; sponsorship is not available. * 8+ years in software or platform engineering, with hands-on production ownership of AI or ML infrastructure. * Has built and operated agentic or LLM systems in production, not only prototyped them. * LLM and agent orchestration with LangGraph, LangChain, LlamaIndex, CrewAI, AutoGen, or the Google Agent Development Kit (ADK). * Model serving and gateway experience (LiteLLM, Portkey, Kong AI Gateway, TrueFoundry, or Vertex AI Model Garden), including routing, fallback, rate limiting, token/usage capture, serving open-weight models (Llama, Mistral, Gemma, Qwen) via vLLM, TGI, Ollama, or Vertex endpoints, and LoRA/QLoRA adapters. * RAG experience with Vertex Vector Search, Pinecone, Weaviate, pgvector, or Elasticsearch. * Evaluation and LLMOps with RAGAS, DeepEval, Promptfoo, or OpenAI Evals, and experiment tracking with MLflow, LangSmith, Langfuse, Arize Phoenix, or Braintrust. * Guardrails and AI security experience with NeMo Guardrails, Guardrails AI, Model Armor, or Lakera. * Observability (OpenTelemetry tracing for LLM and agent calls); Kubernetes and containers; GCP preferred (GKE, Vertex AI, Cloud Run); Terraform, CI/CD, and SRE practice for production AI services. * Strong Python. Preferred Qualifications: * Go or Java for platform services. ## Description ZipStaff is hiring an AI Platform Engineer for our client, a Fortune 10 healthcare and pharmacy leader. You'll build and operate a production agentic AI platform - the layer other engineers build on - and own its reliability, security, and cost. This is a fully remote contract-to-hire role. Candidates must be US citizens or green card holders; sponsorship is not available. Responsibilities: * Build and operate a production agentic AI platform: orchestration, model gateway, agent registry, retrieval, guardrails, evaluation, and observability. * Implement production multi-agent orchestration, including state management and retry/fallback design. * Operate model serving and gateway capabilities, including routing, fallback, rate limiting, token/usage capture, and serving of open-weight models. * Build retrieval-augmented generation, including vector stores, embedding models, chunking, hybrid and semantic search, reranking, and grounding-quality evaluation. * Implement agent infrastructure, including Model Context Protocol (MCP), tool/function-calling contracts, agent registries, permission enforcement, and separation of reasoning from acting agents. * Design evaluation and LLMOps practices, including eval harnesses, offline and online evaluation, regression suites, drift and hallucination detection, and experiment tracking. * Apply guardrails and AI security, including prompt-injection, jailbreak, and data-leakage defense, plus PII detection and redaction in prompts and responses. * Instrument observability for LLM and agent calls (token, latency, and cost telemetry) and quality-signal monitoring that feeds alerting.