Organizing a Conference/Sprint/Workshop

[mrocklin]

We may want to organize an in-person Dask meeting that brings together core developers and sophisticated users. This came up in this month's community meeting and there was general enthusiasm for the idea.

See meeting notes here: https://docs.google.com/document/d/1UqNAP87a56ERH_xkQsS5Q_0PKYybd5Lj2WANy_hRzI0/edit?usp=sharing

Some thoughts from that meeting:

• For our first event this might be somewhat small, maybe 20-40 people.
• It might be something in between a sprint (usually 90% devs 10% users) and a conference (90% users 10% devs)
• After checking in with some people a few of us (@jcrist @mmccarty @kkraus14, @mrocklin) might meet up and discuss logistics

[mrocklin]

To make this concrete I'll propose an hypothetical event below. This is mostly to get people thinking about if this makes sense for them, and if not, why not. Sometimes it's easier to talk about a concrete plan than to create something with full freedom.

So here are some details for our hypothetical event:

• Who: 40 people, a few coming from the pure software side (maintainers at anaconda/nvidia/quansight), a few coming from pure user side (we invite some known users), but most coming from somewhere in between. These are people who support others within their institution, but probably do a bit of maintenance on a project that is relevant to their group, and also relevant to others.

  We do have an open application for talks and attendance, but we arbitrarily decide who comes based on current use, potential use, and diversity (of both personal background and professional domain)

• Where: Let's go for Washington DC. Having this in the US is probably the most likely to start out, but at least DC is easy to fly to, and it's closer than most places for Europeans. We'll probably also get the finance users from NYC (it's an easy train ride) and organizing a venue and hotels in DC is probably easier and significantly cheaper than NYC. We also get US government users, which is nice.

• What: We structure things into talks and then unstructured time in both morning and afternoon. The first day might look something like the following:

  • 8:30: doors open, coffee and light snacks
  • 9:00: welcome and administration
  • 9:15-10: Three 15 minute talks (maybe about common use cases)
  • 10-12: Unstructured time for conversations and discussion
  • 11:45: Quick summary from unstructured time
  • 12: Lunch
  • 1:00-1:30: a few longer talks (maybe about deployment)
  • 1:30-4:00: Unstructured time
  • 4-5: a couple more talks, followed by report from the day.

  Maybe we do this for a couple of days. Or maybe we include more talks on the first day and have more unstructured time on the second.

• When: March. It's soon, but not too soon. It's after the holidays and hopefully DC is starting to warm up.

[mmccarty]

Overall, I think this looks great. Thank you for putting this together. A few things to consider...

• How do the sprints fit into the schedule?
• We could allow folks to propose topics in the morning for breakout sessions focusing on a few of them in the afternoon.
• Perhaps it would be good to get an overview of the road map and key challenges over the next year

[kkraus14]

cc @datametrician here as well

[mrocklin]

OK, I would like to organize a call with a smaller group of organizers to hammer out some short-term logistics. Please 👍 this comment if you would like to be included in this discussion.

[mrocklin]

We're meeting today at 2:15 US Pacific . I'll check the thumbs-up counter above a bit before to see if there is anyone else that I should invite.