Description
We need someone to build production agentic automations, not chatbot demos. Today lead intake, CRM updates, and follow-ups are mostly manual. You’ll turn that into reliable agent workflows with human review where accuracy matters. **What you’ll do** - Build agentic workflows (intake → classify → act → notify) - Connect LLMs to tools via n8n (or Make/Zapier) and APIs - Add RAG where agents need grounded context - Ship a solid MVP, then harden it (retries, logging, guardrails) **MVP** Inbound form/email/Slack → extract + classify → update CRM + Slack summary → queue low-confidence cases for review. **Must-have** - Shipped AI agents or LLM automations in production - Python and/or TypeScript; APIs and async workflows - LangGraph/LangChain, OpenAI/Claude agents, or similar - n8n, Make, or Zapier experience - RAG and vector databases (pgvector, Pinecone, Weaviate, or similar) - Clear communication and pragmatic shipping **Nice to have:** FastAPI/Docker, HubSpot/GHL/Notion/Slack, MCP **Details:** Remote · Hourly · 10–20 hrs/week to start · $40–$75/hr · ASAP **To apply:** Put the word **dubai** at the top of your proposal. Then include 2–3 relevant projects, preferred stack, availability/timezone/rate, and a short MVP approach.