[extropy-chat] RE: [wta-talk] SIAI seeking seed AI programmer candidates

[Adrian Tymes]

--- Eugen Leitl <eugen at leitl.org> wrote:
> On Wed, Jun 02, 2004 at 11:24:24AM -0700, Adrian
> Tymes wrote:
> > Restoring humility: only a single Master's
> > ("computational neuroscience" would be an accurate
> 
> Very interesting, and highly unusual. You definitely
> shouldn't call yourself
> a mere "programmer".  

If I try applying a unique label to myself, I tend to
come off as arrogant.  Not good for obtaining work, no
matter how honest it would be.

> > description; the term UCLA used was
> "biocybernetics"),
> 
> Can you outline the details of the study for us
> who're not comp neurosci
> people? Roughly which areas did the lectures cover,
> and which
> modelling packages/problem sets were used in
> practical work? I'm really
> interested.

It's been almost a decade, so I might forget some of
the details.  If you want a really accurate
description, they're at http://biocyb.cs.ucla.edu/ .
Note that they're using a lot more advanced stuff
since I was there (no surprise, given what's become
available in the intervening years).

> > I've designed ASICs before (that was my first real
> > job, in fact, as an intern one summer during high
> > school), and I've built distributed systems
> designed
> 
> Unfortunately, prototyping ASICs with on-die
> crossbars/switch 
> fabric, multiple ALUs and large on-die memory have a
> very large 
> threshold. This is something requiring serious
> funding to even get started.

Yep - but not as much as you might think.  My employer
had enough funding, but they also had a cheap way to
do it, which is probably why they could afford to have
an intern do the work.  (These days, one can upload
designs to chip fabrication shops and have them run it
for even cheaper.)

> > to run on >1000 nodes (although the dot-bomb I
> built
> 
> SOAPy/XML-RPC stuff, or MPI?

Neither.  Custom design.  Closer to SOAP, I suppose,
but this particular application was never meant to
interface outside of itself, and thus never stood to
benefit from compliance to either standard.

That's actually been an issue I've faced on many
projects.  SOAP and MPI are good for communicating
between applications maintained by different groups,
but they just add overhead if all the applications
involved are maintained by people in the same office -
or who at least can work together that efficiently -
or even by the same person.  I've lost count of the
number of projects I've been on that rolled their own
internal communication protocols for just that reason.
I long ago got to the point where I can almost code a
basic HTTP client (or server) in Perl from memory, and
even that's just used as a minimal overhead protocol
to make things easy to diagnose.

> You've got some extremely unusual background for a
> "programmer", so this
> doesn't really surprise me. Do you know several
> people with roughly your
> background, and are you still keeping in touch?

Nope.  Sorry.