RAG Model Development with UI

I'm looking to build a user-friendly interface similar to ChatGPT, along with a Retrieval-Augmented Generation (RAG) model. This model will process PDF and Word files as input. It's crucial that the model does not hallucinate and only generates responses based on the provided sources. Additionally, I want this project to be open source and hosted with minimal infrastructure requirements.

Requirements:
- User Interface:
- ChatGPT-like design, focusing on ease of use.

- RAG Model:
- Accept PDF, Word, excel and audio as inputs.
- No hallucination; responses strictly from input sources.

- Open Source: Project must be fully open-source.

- Minimal Infrastructure: Cost-effective hosting solution.

Ideal Skills and Experience:
- Expertise in building user interfaces for AI models.
- Experience with RAG models and managing input from various file formats, specifically PDF and Word.
- Familiarity with open-source projects.
- Knowledge of cost-efficient hosting solutions.

Tech Stack Suggestions:
- Frontend: React.js or Vue.js for a responsive UI.
- Backend: Python with FastAPI or Flask.
- RAG Implementation: Use of an LLM like OpenAI's GPT or similar, with a focus on minimizing hallucination.
- Hosting: Consider platforms like Heroku, Vercel, or DigitalOcean for budget-friendly options.

When bidding, you must include otherwise no response:
- Proposed tech stack.
- Chosen LLM.
- Hosting strategy and estimated costs.