Local RAG chatbot for construction business | AI Chatbot

Project Overview I'm a Civil Engineer at a material testing company in Saudi Arabia. My team of fresh graduates needs quick access to technical information on construction material testing, QA/QC, geotechnical logging, site inspections, and report preparation. I want a local chatbot prototype that answers questions based on our uploaded documents (PDFs, Word, TXT). The app must run on our Windows laptops with no cloud deployment. This is a prototype/demo phase. I need a working app I can run locally and present to my team for feedback. If approved, I'll hire the same freelancer to build the full production version with a significantly larger budget. I'm paying for the Google Gemini API and will provide the key. What I Need Built: Document Upload & Processing Users can upload PDF, DOCX, and TXT files Documents are processed, chunked, and embedded for retrieval Support for adding new documents after initial setup Chat Interface Simple, clean web-based UI (functional is fine — no fancy design needed) Users type questions and get answers based only on uploaded documents Answers should cite which document(s) the information came from Local-First Architecture Runs entirely on local machines (Windows laptops) No cloud hosting or external vector database required at this time. Use local vector store like FAISS or ChromaDB I will provide the Gemini API key Setup & Handover: Clear, step-by-step instructions on how to install and run the app on our 6 laptops Include a README.md with setup instructions Provide a requirements file and a simple install script A video tutorial explaining how to get the app running on my laptops Tech Stack Preferences: LLM: Google Gemini API (Gemini 2.0 Flash or similar) Framework: Python (FastAPI/Flask) or Node.js — open to freelancer's recommendation Vector Store: Local-only (FAISS, ChromaDB, or in-memory store) Embeddings: Gemini Embeddings or Sentence Transformers Frontend: Simple HTML/JS or lightweight React — no complex UI needed for prototype Document Parsing: PyPDF2, python-docx, or similar I'm open to the freelancer's tech stack recommendation if it meets the local-first requirement and uses the Gemini API. Deliverables: Source code in a clean, well-organized repository Working prototype I can run locally on my laptop Setup documentation with: Prerequisites (Python version, etc.) How to install dependencies How to configure the Gemini API key How to run the app How to upload documents How to ask questions Video tutorial (screen recording with voiceover, 15–30 min) walking me through: How to install and set up the app on a fresh Windows laptop How to configure the Gemini API key How to upload documents and start asking questions How to add new documents after initial setup A live 30-min video call walkthrough where I can ask questions What I Will Provide Google Gemini API key Sample technical documents for testing Prompt feedback during development Clear requirements and quick responses What I'm Looking For in a Freelancer Experience building RAG (Retrieval-Augmented Generation) applications Familiarity with Google Gemini API or similar LLM APIs Comfortable building local-first apps (no cloud deployment) Good communication skills — must explain technical setup clearly Willingness to work within a prototype budget with clear understanding that full production work will follow at a higher rate Ability to explain technical setup in plain English for non-technical users Bonus: Experience with document-heavy or engineering/technical domains Timeline Timeline: 1–2 weeks for prototype delivery Future Work: If the prototype is well-received by my team, I'll expand the scope for a full production version with a significantly higher budget ($1,500+), including features like user authentication, enhanced UI, multi-user support, and advanced document management. To Apply, Please Include: Links to similar RAG/chatbot projects you've built (GitHub, demos, etc.) Your preferred tech stack for this project and why Estimated hours/days to deliver the prototype Your availability — can you start immediately? Confirm you can provide a video tutorial explaining setup and usage Any questions you have about the requirements Add the word "Playable" at the top of your request so i know you have read the entire description.