Backend For Pipelines

• generate code for ML tasks using templates
  • tasks✔️
  • data processing✔️
  • model definition✔️
  • training configuration✔️
• communicate with UI using FastAPI
  • receive config✔️
  • send back generated code✔️
  • send training command, inference results
• run training or inference scripts and return result to UI

Running the server

• install uv with brew install uv or python -m pip install uv

• sync dependencies uv sync from root folder of repo

• activate venv source .venv/bin/activate (/activate.<shell_ext> if present in .venv/bin)

• run server python modules/main.py

• check api at http://127.0.0.1:8000/docs

• or

• run the release ./main

TODO: Build into a cross platform binary

• build one using pyinstaller, need to build on other OS ✔️
• Need to bundle witht The UI using tauri side-car

TODO: Pending functionalities

1. run the generated code using the UI instead of needing to manually save the file and run
2. Test the server, currently the App only builds something that is basically a boilerplate builder,

• • need to test with different configs, datasets, and actually building the dataloaders.

3. convert the fastAPI app into rust based. there isnt much benefit of using python here.