AI Support Agent + RAG Chatbot for a DTC Support SaaS

We're building an AI support layer for DTC e-commerce brands, and we're hiring for a focused one-week sprint. Scope is a multi-tenant SaaS shell with a RAG pipeline over each brand's own content, plus an AI agent layer with tool use and function calling. On the retrieval side that means document ingestion, chunking, and embeddings into a vector database (Pinecone or pgvector) with semantic search, then grounded LLM answers with guardrails so the agent never hallucinates a policy. By the end of the week we want the core loop running on sample data: grounded RAG answers from uploaded docs, one working agent action through a real tool call against test data, clean escalation to a human with full conversation context, and a basic chat UI we can click through. Tech we lean toward: Python and FastAPI on the backend, Node, React or Next.js on the front, Postgres or Supabase for data, LangChain for orchestration, and OpenAI or Claude behind a model-agnostic layer so we're not locked to one provider. We want your reasoning on the stack and the prompt engineering. If the initial work lands, it turns into a full SaaS build, so we care about clean, production-minded code, not throwaway demo scripts.