Senior AI Engineer needed for AI Agents, LLM, and RAG for Real Estate multi-tenant SaaS application

We run a Real Estate management SaaS used by hundreds of companies. Our platform follows a multi-tenant architecture, where each firm operates with its own cases, clients, and documents We’ve been live for about four years, and the one thing our customers keep asking for now is AI. We’re looking for a senior AI engineer who has built AI agents, large language model (LLM) applications, and RAG systems that run in production with real users. We care much more about that than about how many frameworks someone has listed. The first part of the project is a retrieval-augmented generation (RAG) assistant over each customer’s own documents. Our customers upload leases, HOA rules, inspection reports, and vendor contracts, and they want to ask things like “which leases expire in the next 60 days?” and get a correct answer with a link to the source. If the answer isn’t in their documents, it should say so. Some files are scanned PDFs, and some data we pull from vendor and county websites, so experience with OCR and data scraping pipelines would help a lot here. All of it needs to end up clean in the database The second part is AI agent development. We have a full REST API already, and we want LLM-powered agents that use it as tools: creating maintenance tickets from a tenant’s message, checking vendor schedules, drafting notices, and building monthly owner reports. We’ve used LangChain a little and like it, but we’re open if you think LangGraph or a lighter custom setup is better. The part we care about most is tenant isolation. Data from one company can never leak into another company’s results, whether through vector search, caching, chat memory or logs. If you’ve built multi-tenant SaaS before and solved this, please tell us how. We’d also like to track AI usage and cost per tenant (we’ll charge for AI as an add-on), route simple questions to cheaper models, and have a simple evaluation setup so we know when answers get better or worse. Our stack is Python with FastAPI on the backend (a couple of older internal tools are in Django), and a Next.js and React frontend. Everything runs on AWS. Because the AI features touch the backend, the API, and the UI, we'd ideally want a full-stack developer who can work across all of it, rather than someone who only does the AI part. You’d mostly work with our team and me on in-house developers. We’ll start with a paid two-week trial. We need at least 4 hours of overlap with US Eastern time. When you apply, please tell us: How would you make sure one tenant’s data never shows up in another tenant’s answers? An AI agent, LLM, or RAG project you built that’s in production, and one thing that went wrong with it. Which framework you’d use for the agents, and why. Please no copy-paste proposals. We read every one. Thanks,