Portfolio Project
Chatbot (LoRA + RAG)
RAG Chatbot Fine-Tuned with LoRA
Context
Generic chatbots lacked Visit Grand Junction's tone and rarely suggested our content.
Role
- Led the end-to-end prototype: data ingestion, retrieval, fine-tuning, deployment, and web integration.
Approach
- Scraped Visit Grand Junction pages and created a FAISS retrieval index.
- Automated a fine-tuning dataset with GPT-OSS 20B via Ollama.
- Fine-tuned Mistral 7B on the generated QA set and deployed it to AWS SageMaker.
- Created Lambda endpoints so the website can interact with the model.
Impact
- Serverless RAG chatbot scales on demand and returns grounded answers with references after a 10‑minute server warm‑up.
Links
Notes
Built from public Visit Grand Junction web content; tuned to return grounded answers with references.