Pre-submission enquiry

[jacopotagliabue]

Dear Editors,

We wish to propose a submission for Metaflow (https://github.com/Netflix/metaflow), a popular open source tool for machine learning workloads.

The framework seems to comply with most of the requirements in terms of "constituting a substantial body of work" and openness. In our view, Metaflow does indeed "supports the (...) the execution of research experiments" - as such, it has already been cited in several peer reviewed publications, such as Nature journals (https://www.nature.com/articles/s41598-022-23052-9) and top-tier ML venues (SIGIR: https://arxiv.org/abs/2107.03256, RecSys: https://arxiv.org/pdf/2107.07346.pdf).

However, we do recognize the primary purpose of the tool is not to provide new ways to perform scientific computation, but instead making computations scalable, repeatable, reliable, self-documented, without assuming any infrastructure knowledge from the part of the scientist.

For this reason, we are unsure if our submission, centered on usability, abstraction and scale, would be a good fit for the scope of the journal, and we wish to consult with you on what you deem interesting and appropriate.

Thank you in advance for your kind attention.

Best,

JT

[arfon]

Thanks JT. Let me check with the wider editorial team and get back to you about this.

[arfon]

@jacopotagliabue – we'd welcome a submission to JOSS for metaflow.