combodock

• 

  combodock-architecture

• 

  codeset-docker-compose-json-bot

• 

  codeset-dockerfile-json-bot

• 

  combodock-docker-compose-up-build

• 

  combodock-run

Inspiration

I am interested in generative AI and also with docker tools, hence I have proposed to submit a project combining these technologies for this Challenge

What it does

It allows users to ingest, then chat and search using its own Custom Json data

How we built it

I have studied the public genai-stack (neo4j, docker,...) archive and I have decided to add a new module enabling to import its own Custom JSON data and request interactive query about this own JSON data.

• use streamlit to enable users to select interactively and query multiple Json data files after automatic ingestion.

• add specific json-bot.Dockerfile to allow containerization and deployment of the Json processing application

• update docker-compose.yaml file to reference this new json-bot module and dockerfile

Challenges we ran into

• JSON data load and integration with neo4j

• deploy and run genai-stack on windows 10 docker > fixed by using WSL2 Debian environment and updating the docker version

Accomplishments that we're proud of

• Minimum Viable Product of json bot module, to allow end users to upload their own specific Json data as multiple files and query interactively this specific json data

What we learned

• usage of genai-stack project in different contexts

• new way to integrate module custom Dockerfile inside docker-compose.yaml

What's next for combodock

• make the json bot app available remotely to other users

• allow integration of new repo data in order to query about them later on

Built With

• docker
• docker-compose
• genai-stack
• langchain
• neo4j
• ollama
• openai
• python
• streamlit