AI Chatbot Developer Needed – RAG-Based Customer Support Assistant (OpenAI/Claude + Next.js)

We’re a mid-sized e-commerce company with about 2,000 support tickets a month. Most are repetitive questions about orders, returns, and product details. We want an AI assistant that answers these automatically using our own knowledge base, and hands off to a human agent when it can’t. What we need: A chatbot widget embedded on our Next.js website Retrieval-augmented generation (RAG) over our help docs, FAQs and product catalog (around 500 pages) Integration with our order system through a REST API, so the bot can check order status Human handoff to our Freshdesk account when confidence is low A simple admin panel to upload or update knowledge base documents Basic analytics: questions asked, resolution rate, handoff rate Tech preferences: LLM: OpenAI or Claude (open to your recommendation) Vector DB: Pinecone, Supabase pgvector or similar Backend: Node.js or Python Deliverables: Working chatbot deployed on our staging site Source code in our GitHub repo Documentation for updating the knowledge base Two weeks of post-launch support Ideal freelancer: Has built and shipped at least two RAG or LLM-based applications Understands prompt engineering, hallucination control and evaluation Communicates clearly and gives regular progress updates To apply, please answer: Share a link or demo of an AI chatbot or RAG system you’ve built. How would you prevent the bot from giving incorrect answers about our policies? Which LLM and vector database would you recommend for this, and why? What’s your estimated timeline?