Über
Fixed Price | Estimated 15–20 hours | Start Immediately
Project Overview
We are property insurance claims experts building an internal AI knowledge base of over 45,000 client documents. Staff need to query this data in plain English and get answers, analysis, and insights grounded in our actual files — not just generic AI responses from internet queries.
The system combines three data layers: our 45K document library, Claude's base knowledge, and live web results — all returned in a single answer.
Tech Stack (Non-Negotiable)
• AnythingLLM or Kotaemon — pre-built RAG system via Docker
• Claude API (Anthropic) — intelligence and chat layer with web search enabled
• dedicated server
• JobNimbus API — document export script (AI-assisted, Cursor or Windsurf)
• n8n — weekly sync automation
Scope of Work
• Verify JobNimbus API supports bulk document attachment export
• Provision and configure server with Docker
• Deploy AnythingLLM or Kotaemon and configure for 45K document scale
• Build JobNimbus export script using AI coding tools — pulls all document attachments into RAG intake folder
• Run full overnight export and verify completion
• Connect RAG system to Claude API with web search enabled
• Test with real business queries, tune chunking for construction/insurance document types
• Set up staff browser access with basic user authentication
• Configure n8n weekly sync to keep documents current automatically
• Basic server monitoring and alert setup
What We Are NOT Building
• Custom UI or chat interface — staff use Claude directly
• Custom OCR pipeline — handled by the RAG tool out of the box
• Enterprise-grade monitoring or audit infrastructure
• Multi-tenant or per-customer data isolation — internal staff only
Developer Requirements
• Proven experience deploying RAG systems (AnythingLLM, Kotaemon, LangChain, or equivalent)
• Comfortable with AI-assisted development — Cursor or Windsurf is expected, not optional
• Experience with Claude API or OpenAI API integrations
• Able to work with Hetzner/Linux server environments
• Can show prior RAG or LLM integration work — portfolio or GitHub required
Budget & Timeline
Budget: $1,500–$2,000 fixed price. Timeline: functional MVP within 2 weeks of start.
This project is scoped for AI-assisted development. We expect AI tools to handle boilerplate — your value is architecture judgment, integration, and testing. Please do not apply if you plan to build this manually from scratch and charge accordingly.
To Apply
Send a brief note covering: (1) a RAG or LLM integration you have shipped, (2) your experience with AnythingLLM or Kotaemon specifically, and (3) your honest read on the biggest technical risk in this project.
Contract duration of 1 to 3 months.
Mandatory skills: Artificial Intelligence, Machine Learning, claude api, rag pipeline, n8n, Python, jobnimbus api, Linux
Sprachkenntnisse
- English
Hinweis für Nutzer
Dieses Stellenangebot stammt von einer Partnerplattform von TieTalent. Klicken Sie auf „Jetzt Bewerben“, um Ihre Bewerbung direkt auf deren Website einzureichen.