Site RAG AI Agent Integration
Budget: $10 – $55 USD
I already have a retrieval-augmented-generation (RAG) agent sitting in a public GitHub repo, written in Python. Your job is to take that code base, wire it up to my existing website, and make the assistant available to visitors so they can:
My agent code is @ https://github.com/eternityready2/eternityrag
• find and display relevant content from my site and other approved sources
• hold live discussions with the user
• provide thoughtful, Biblically grounded insights when the conversation turns to Christian topics
• scale gracefully so I can add additional skills later
BACKEND CONSTRAINTS
My priority is to keep monthly spend close to zero. I am therefore leaning toward open-source or very low-cost models (e.g. Llama 3, Mistral, Ollama, LocalAI) running on my own VPS or a lightweight managed service. If you can demonstrate that a small-tier commercial plan (Gemini Pro Lite, OpenAI 3.5 free quota, etc.) fits the cost target, I am open to it, but the default should be free/open.
TECH STACK FACTS
• Current agent: Python, LangChain-style architecture
• Repo includes basic vector-store retrieval but no production DB
• My site: simple Flask back-end with a vanilla HTML/JS front end (can be adjusted)
• Hosting: Ubuntu server with SSH access and Docker installed
WHAT I NEED FROM YOU
1. Review the GitHub code and confirm what, if anything, must be refactored for production.
2. Select and configure an appropriate model backend; document RAM/CPU/GPU needs and any licensing notes.
3. Build the web integration: an embedded chat widget or full-page chat route, whichever is lighter.
4. Wire up retrieval so the agent can pull content from my site’s articles and future uploads.
5. Deploy to my server, run a short load test, and leave me with a one-command update script plus clear README instructions.
ACCEPTANCE CRITERIA
• Chat loads in <2 s and answers within 5 s on a cold start.
• Queries like “Show me yesterday’s devotional” return the correct page link.
• Discussion remains on-topic for at least five exchanges without hallucinating doctrine.
• Total monthly model cost ≤ $10 (preferably $0).
If you have solid Python, AI model ops, and web-integration experience, I look forward to your proposal and a brief outline of which backend you would choose and why.
The project budget will be a max of $55 usd. I only need a cloud ai engine attached to my code and then my code deployed and added to my website. There will be some web development and coding involved.
My agent code is @ https://github.com/eternityready2/eternityrag
• find and display relevant content from my site and other approved sources
• hold live discussions with the user
• provide thoughtful, Biblically grounded insights when the conversation turns to Christian topics
• scale gracefully so I can add additional skills later
BACKEND CONSTRAINTS
My priority is to keep monthly spend close to zero. I am therefore leaning toward open-source or very low-cost models (e.g. Llama 3, Mistral, Ollama, LocalAI) running on my own VPS or a lightweight managed service. If you can demonstrate that a small-tier commercial plan (Gemini Pro Lite, OpenAI 3.5 free quota, etc.) fits the cost target, I am open to it, but the default should be free/open.
TECH STACK FACTS
• Current agent: Python, LangChain-style architecture
• Repo includes basic vector-store retrieval but no production DB
• My site: simple Flask back-end with a vanilla HTML/JS front end (can be adjusted)
• Hosting: Ubuntu server with SSH access and Docker installed
WHAT I NEED FROM YOU
1. Review the GitHub code and confirm what, if anything, must be refactored for production.
2. Select and configure an appropriate model backend; document RAM/CPU/GPU needs and any licensing notes.
3. Build the web integration: an embedded chat widget or full-page chat route, whichever is lighter.
4. Wire up retrieval so the agent can pull content from my site’s articles and future uploads.
5. Deploy to my server, run a short load test, and leave me with a one-command update script plus clear README instructions.
ACCEPTANCE CRITERIA
• Chat loads in <2 s and answers within 5 s on a cold start.
• Queries like “Show me yesterday’s devotional” return the correct page link.
• Discussion remains on-topic for at least five exchanges without hallucinating doctrine.
• Total monthly model cost ≤ $10 (preferably $0).
If you have solid Python, AI model ops, and web-integration experience, I look forward to your proposal and a brief outline of which backend you would choose and why.
The project budget will be a max of $55 usd. I only need a cloud ai engine attached to my code and then my code deployed and added to my website. There will be some web development and coding involved.