For my workshop I developed a prototype for a small AI project. The idea was that students set up a virtual machine with VirtualBox running Debian and then build their own small AI web apps using Python (vibecoding) and Flask.
The task was to choose three out of five simple AI models and make them usable through a self-built (vibecoding) web interface (HTML / CSS / JS).
The available models were:
- BLIP, image captioning
- DistilBERT, sentiment analysis
- GPT-2 (117M), text generation
- EasyOCR, text recognition in images
- Real-ESRGAN, image upscaling
I deliberately selected the models so they could run on typical school hardware. As a benchmark I used my old notebook (Intel Celeron N3150, 4×1.6 GHz, 4 GB RAM).
The goal was not only to "use AI", but to show how models can be integrated, how web interfaces are built, and how a small system can be assembled from different components.
(More information can be found in the comments.)
The video shows the prototype of the project.
Background music: Eiffel 65 - Blue (Da Ba Dee) (Rock Version by Kenke) https://www.youtube.com/watch?v=A-M5gMmDRJ8
#AI #Python #Flask #FOSS #OpenSource #FreeSoftware #Linux #Debian #Privacy #DigitalSovereignty #Fediverse #DigitalLiteracy #MediaLiteracy #TechEducation #ComputerScience



