Remote AI OS Architect — 6-Month Contract

Harvey Nash; United States; Sur place; USD 117 000 - 138 000; Plein temps

Résumé du poste

Harvey Nash seeks an experienced AI OS Engineer for a high-impact, 6-month contract. You will lead the architecture and integration of our next-generation AI Operating System—a core orchestration framework designed to manage autonomous agents, multi-LLM routing, context memory systems, tool execution, and cloud compute pipelines.

As a contract-focused role, you will turn architectural blueprints into production-grade infrastructure, implement evaluation frameworks, and optimize latency and

Qualifications

• 5+ years of production software engineering experience with 2+ years in agentic frameworks or LLM infrastructure.
• Proficiency in Python and TypeScript/Node.js; strong async execution knowledge.
• Experience with vector databases and hybrid search.
• Hands-on with Docker, Kubernetes, and cloud platforms.
• Knowledge of REST/gRPC APIs and microservices architecture.

Responsabilités

• Design and deploy agentic workflows and runtime environments for internal AI apps.
• Implement robust retrieval systems and vector databases for context management.
• Architect multi-model routing layers for cost-efficiency and low latency.
• Develop secure sandboxes for tool execution and AI safety protocols.
• Build evaluation harnesses to monitor model drift and latency.

Connaissances

Python

TypeScript/Node.js

async execution

agent architectures

LLM infrastructure

Outils

Docker

Kubernetes

vLLM/Triton

LangGraph

Qdrant/Pinecone

Description du poste

Harvey Nash seeks an experienced AI OS Engineer for a high-impact, 6-month contract. You will lead the architecture and integration of our next-generation AI Operating System—a core orchestration framework designed to manage autonomous agents, multi-LLM routing, context memory systems, tool execution, and cloud compute pipelines.

As a contract-focused role, you will turn architectural blueprints into production-grade infrastructure, implement evaluation frameworks, and optimize latency and