Help us help you: Survey [<4 mins] to inform Jupyter extension development (submit before 10/15)

[fern...@hypernetwork.io]

Hey all!

My team and I are working on a Jupyter extension that allows Jupyter users running compute-heavy code (i.e. ML training) to drastically reduce total cumulative runtime of their projects. How? By first providing visibility on a dashboard into their trusted peers’ idle compute power (labmates’ / friends’ laptops & desktops) and second, sending the code to this other laptop and commanding the execution via our extension.

We'd love to hear from the community via this survey to assess a) if the pain point is big and frequent enough to warrant our extension and if so, b) specific features you'd like to see in the extension. Please submit before 10/15.

We're trying to reach every loyal Jupyter user out there so my apologies if you hear from us in another forum over the course of the weekend; thank you so much for your time and thoughtful responses.

- (A different, less famous) Fernando

[fern...@hypernetwork.io]

Hey all, happy Friday!

Thank you so much for your responses to date. Having conducted surveys for other matters in past lives and knowing firsthand that they are not anyone's favorite thing to do, I was surprised by the engagement the Jupyter community showed. But then again, knowing what I knew about this tight-knit community of engaged users, was I surprised?

In addition to thanking those who have responded, today I check in to remind everyone that the survey will be taking responses for 10 more days (10/15). 4 minutes of your time, 16 questions to inform the development of a useful Jupyter companion.

Thank you for your time!

Fernando

[Simon Biggs]

Can I recommend not just looking into a dashboard and python library, but maybe a custom JupyterLab kernel based on the python3 kernel might be the way to go.

Potentially users can program just as they are used to and whenever a library is called that is supported by hypernet then that is detected and handled by the kernel to also run through your distributed platform.

Making it so people can use numpy, tensorflow, etc just as they're used to would be quite the boon.

[fern...@hypernetwork.io]

Hey all, this is just a final friendly reminder. We'll be taking survey responses until 11:59pm tomorrow Tues. 10/16, Pacific Time. Again, many thanks for the thoughtful responses to date. Hope everyone has an amazing week.

- Fernando