SupraLabs_

AxionLab does not have a GPU. Not a 5090. Not a 4090. Not even a polite little 3060. Just a fairly bad iGPU that treats real training like a personal insult 😂

That is a problem when you want to train models, not slideshows.

Kaggle was the workaround. It was also the joke.

Two T4s. Free-ish. Fine until it isn't. Kernels dying mid-run. Sessions vanishing. Crashes that feel personal. The classic Kaggle tax: wait, retry, pray, lose the notebook anyway 😭

AxionLab was done being a Kaggle speedrun.

Then a sponsor showed up

We found a sponsor for Runpod. Real GPUs. Real persistence. No more begging a free notebook not to explode.

We also founded a small team around it. Not a corporation. A lab that can actually press start and mean it.

We built a tiny dashboard

Not a platform. Not a product launch. A small internal dashboard to start and manage pods without living in fifteen browser tabs.

One place to spin pods up. One place to shut them down. Direct jump into JupyterLab. Direct jump into the bash terminal. That is the whole interface. Dead simple. Exactly what we needed.

Visit it at https://lh-tech.de/runpod/deploy.html

AxionLab is happy now

No more crashing kernels. No more mystery Kaggle deaths. No more two T4s pretending to be a lab 😂😂😂

Just a pod, a notebook, a terminal, and work that actually stays alive.

Small team. Sponsored Runpod. Tiny dashboard. Finally training like we meant it.