Thank you to everyone that has contributed to this conversation. I've gotten some leads and, more importantly, a better idea of cost of good instruction with proper materials, etc. I'll report back if we find any novel solution.
As for experience developers, they learn in a variety of ways and a few focussed days can significantly speed up the learning process for most of them. Participation with other developers (in particular co-workers), jointly asking an expert questions and working through exercises enhances the learning. Certainly not expecting anyone to be an expert by the end of a week of any kind of training. Yes, it can all be learned with online tutorials, or splitting up topics to focus on and lecturing to the group or whatever, just slower and potentially missing a bigger picture (like learning the Python way). Chad On Jul 17, 2013, at 4:57 PM, Matt Stumbaugh <[email protected]> wrote: > Very interesting conversation. In a perfect world, it seems like a > seminar/workshop at UW could be setup to meet the needs of Chad's group and > those of many other groups-- because while the topics are specialized, they > certainly aren't unique to seismology. Such a format wouldn't put the burden > on one person, a lesson on any given topic would be much more likely to be > well prepared, and it would bring like-minds together. Wrangling such a > mechanism together probably wouldn't be trivial but it seems to be in the > vein of academia and it seems to me like it would leverage scarce resources. > An alternative might be to try to find an additional interested group to > pitch in and hopefully alleviate the sticker shock (and provide the requested > coverage/format). > > > > > On Wed, Jul 17, 2013 at 3:17 PM, Chris Barker - NOAA Federal > <[email protected]> wrote: > On Wed, Jul 17, 2013 at 10:08 AM, Chris Calloway <[email protected]> wrote: > > > Learn Numpy/SciPy/Matplotlib applied to your domain in three days? I don't > > think so. Maybe get an overview in three days. Heck, it can take someone > > with gobs of experience three days just to optimize your Fortran compiler or > > graphics rendering engine for good performance with Python's scientific > > stack. > > or far more trivial things, like getting > numpy/scipy/matplotlib/hdf/netcdf/mayvi? all installed an working -- a > good reason why people buy EPD or Anacoda. > > > So my other observation is going towards a deeper issue. If you are sending > > "experienced developers" to advanced stack trainings, you are doing it > > wrong. Experienced developers learn by exploring and experimenting for > > themselves. > > Well, yes, and that's what is going to happen over the > weeks/months/years is takes to really become proficient. However, a > focused few days could give folks a pretty good jumpstart. > > > My best > > recommendation for getting a shop of experienced developers up to speed on > > the Python scientific stack in the most accelerated way possible is to hire > > a developer or two with that experience and stick them in your bull pen with > > your existing developers: mentoring and especially pairing are way > > underrated. > > Great idea -- but it would take culture ans structure, too -- in my > shop, new folks are coming up to speed mostly on their own, and, I > think, not benefiting as much as they could from, let's call it > mentoring, because: > > a) they don't ask for help as readily as they might. > b) the more experienced folks may not see (and their job description > may not) their job as trainers/mentors. > c) we don't enforce code review (we should!) > > Another idea is to hire a consultant to come in periodically and > consult and code review -- people present a problem, and they get help > working out how to get started solving it. > > Though maybe $10k or less wont et you far that way either... > > I've also found that there is a bit of barrier to entry, particularly > for the less experience developers -- why take the time to figure out > how to write a little python program when you can get the job done by > hand with a spreadsheet? Some focused time to get up to speed can > really help here. > > > I would only recommend > > PyCamp to "experienced developers" if they are having difficulty with doing > > things the Python way or want a quick immersion in Python culture. > > That's another reason to do some training up front, rather than leave > people ot figure it out themselves... > > -Chris > > > -- > > Christopher Barker, Ph.D. > Oceanographer > > Emergency Response Division > NOAA/NOS/OR&R (206) 526-6959 voice > 7600 Sand Point Way NE (206) 526-6329 fax > Seattle, WA 98115 (206) 526-6317 main reception > > [email protected] >
