Portfolio Project

Chatbot (LoRA + RAG)

RAG Chatbot Fine-Tuned with LoRA

Machine Learning Automation Python Ollama AWS Docker

Ask a question and get an answer grounded in the site content (with citations).

  • Click a quick prompt, or type your own question in the chat box.
  • Press Enter or click “Send” to submit.
  • Use the citations in the response to jump to the referenced pages.
  • If the service is warming up after idle time, wait for status to turn ready and try again.

STAR Summary

Situation
Off-the-shelf chatbots didn't sound like Visit Grand Junction, and they rarely pointed people to our pages.
Task
Built the prototype from crawl to deployment, including the web demo.
Action
  • Crawled Visit Grand Junction pages and built a FAISS retrieval index.
  • Generated a fine-tuning dataset by prompting an open-source LLM locally (via Ollama) to create Q&A pairs from the corpus.
  • Fine-tuned Mistral 7B with LoRA on the Q&A set and deployed it to AWS SageMaker.
Result
  • The demo answers with citations; the first request after downtime can be slower due to cold-starting the backend.

Notes

Uses public Visit Grand Junction pages. Answers include citations.