Back-End / AI Engineer - Full-time - Worldwide

About the company

UHP is a veteran-focused training institution near Bentonville, AR, building infrastructure to transform technical education delivery at scale. We're launching a multi-agent AI platform with a hard July 13, 2026 production deadline.

The Role

Own the AI and infrastructure layer: multi-agent orchestration on Anthropic Claude API, RAG pipelines on Supabase + pgvector, agent memory and audit logging, tool calling architecture, and our self-hosted FFmpeg video pipeline (ClipForge).

Stack: Next.js 14 API routes, TypeScript, Supabase (Postgres, RLS, pgvector), Railway, Anthropic Claude API, FFmpeg, Resend, Twilio.

What You'll Do

• Design and implement multi-agent workflows with Claude API: coordination, tool calling, memory, audit logging
• Build RAG pipelines with pgvector: embeddings, vector search, data ingestion
• Develop scalable Next.js 14 APIs with proper auth, rate limiting, RLS policies
• Maintain ClipForge video rendering pipeline with queue management and error handling
• Harden infrastructure under load: monitoring, alerting, circuit breakers, graceful degradation
• Integrate Resend (email) and Twilio (SMS) for notifications

Required Experience

Must Have:

• Shipped production AI systems where LLMs make tool calls, retrieve from vector stores, and operate in business logic at scale (not research, not just prompt engineering)
• 5+ years senior backend engineering: APIs, databases, infrastructure
• Expert TypeScript/Node.js
• Deep PostgreSQL knowledge: query optimization, RLS
• Hands-on vector database experience (pgvector, Pinecone, Weaviate, Qdrant)
• Direct experience with Anthropic Claude API, OpenAI API, or similar LLM APIs

Strongly Preferred:

• Next.js 14 API routes
• FFmpeg or video processing
• Queue systems (BullMQ, Inngest)
• Railway or similar PaaS
• Production scaling experience