Ana Ndongo·Oct 19Opening Jupyter with GPU like opening a browser tabThis is how the full workflow looks with Datoki.
Inlanavajasuiza.techbyAna Ndongo·Sep 28What if the cloud wasn’t the only option?AWS. Azure. Google Cloud.
Inlanavajasuiza.techbyAna Ndongo·Sep 7How to run heavy notebooks without setting up cloud infrastructureThe specific problem
Inlanavajasuiza.techbyAna Ndongo·Aug 17Streamlit is not deployment. The difference between a demo and a real productStreamlit is a brilliant tool.
Inlanavajasuiza.techbyAna Ndongo·Jul 27When setting up the environment takes longer than the experimentThere’s a very specific moment that almost everyone working with notebooks has hit.
Inlanavajasuiza.techbyAna Ndongo·Jul 6Your model in production is already wrong. Retraining, drift and continuous evaluationThe day you deployed your model, it worked well. The metrics looked solid. You put it into production and moved on.
Inlanavajasuiza.techbyAna Ndongo·Jun 24Nobody Needs a GPU All DayMost of the time, your GPU is waiting for you.
Inlanavajasuiza.techbyAna Ndongo·Jun 17The workflow no one talks about: local for daily work, GPU only when you need itThere’s a way of working with notebooks that, once you try it, is hard to unsee.
Inlanavajasuiza.techbyAna Ndongo·Jun 10What Makes Blockchain Interesting for Data Scientists?Public data, network analysis and the opportunity to study behaviour at scale.
Inlanavajasuiza.techbyAna Ndongo·Jun 3CUDA Out Of Memory: Why torch.cuda.is_available() Wasn't Enough (And How I Fixed It)While testing a notebook on Datoki, I ran into a problem that seemed impossible at first.