Shivangx01b / pdf-chatbot-task

pdf-chatbot-task

Production level api to read RAG your PDFs using LlamaIndex, LangChain and OpenAI

• Uses Redis as broker
• Uses Celery for task manager
• Uses async fastapi for accepting "n" numbers of requests

How to Run

a) Docker build

• Step 1 docker build -t pdf_rag .

• Step 2 docker run -it -e OPENAI_API_KEY="<your openai key>" -e slackchannel="<your slack channel id>" -e slacktoken="<your slacktoken" -p 8000:8000 pdf_rag

b) Send request

• Step 1 First upload the pdf once api is up and running

  curl --location 'http://localhost:8000/upload' \
  --form 'file=@"/C:/Users/shivang/Downloads/handbook.pdf"'
  

• Step 2 Ask your questions with the taskid received from "Step 1"

  curl --location 'http://localhost:8000/query' \
  --header 'Content-Type: application/json' \
  --data '{"taskid": "2580a111-8437-4e65-b483-0769e50b3a1f", "question": "What is shivang in Zania Inc"}'
  

How to improve Code

• Can use cache to save query cost when agent is called to check the response from RAG
• Can use multiple agents as a feed back loop for less wrong answers
• Can use go api or rust api for blazing fast api response in production