CPR LLM Research Assistant

I developed a research assistant chatbot utilizing Google Cloud infrastructure. The project involved several key components:

RAG System Development:
I created a Retrieval-Augmented Generation (RAG) system using Google Cloud's Vertex AI. This system efficiently processes and retrieves relevant information from a large corpus of scientific literature from the Spinal Cord Injury Research Evidence (SCIRE) Professional database and the Purdue Center for Paralysis Research.
Integration with Dialogflow CX:
To enhance the chatbot's conversational abilities, I integrated the RAG system with Dialogflow CX. This integration allows for natural language understanding and generation, creating a more intuitive user experience.
Custom Knowledge Base:
The chatbot utilizes over 80 SCIRE toolkits and treatment guidelines supplemented by 150 research papers from the Center for Paralysis research, creating a specialized knowledge base in our field of study.
Gemini 1.5 Flash API Implementation:
I leveraged the Gemini 1.5 Flash API to power the chatbot's responses, ensuring rapid and accurate information retrieval and generation.
Deployment:
The final step involved deploying the chatbot as a fully functional clinical research assistant, capable of answering questions, identifying relevant papers and providing treatment suggestions based on our lab's extensive research.