SKILLGAPPed

Hackaton project

Build

DEFAULT PORT: 3000

To use server you need to:

1. Have node.js installed
2. Have npm installed
3. Installed pytorch (with cuda if you want to use GPU)

Pre-startup, do it only one time after download/update (if dependencies change):

1. download the source code

2. run command:

   2.1 Linux: "npm run BuildProjectLinux" for GPU usage, "npm run BuildProjectLinuxNoCuda" for CPU usage

   2.2 Windows: "npm run BuildProjectWindows" for GPU usage, "npm run BuildProjectWindowsNoCuda" for CPU usage,

startup:

1. deployment

   1.1 inside project directory "node index.js" to start server without live updates

2. development

   2.1 programming backend: "npm run devStart" to start server with live updates (backend only)

   2.2 for building front end there is command "npm run Build"

   2.3 programming frontend: "cd front_end" and "npm run dev" (for working with only front end you dont need to build app, but for it to work with backend YOU WILL NEED to do that)

Infrastructure

Backend: Node.js with Express.js

Frontend: React and Vite

AI pipeline:

1. python script "Kowalski" which takes dataset, fixes it and generates prompt for LLM

2. LLM (using llama.cpp) which takes the prompt and generates .bumbuojam (pseudo) code for the interpreter

3. python script "Interpreter" takes the .bumbuojam code and generates valid plotly vizualizations

4. server sends those vizualizations back to user