Y Combinator | Software Engineer | San Francisco, CA | ONSITE | Full-time

You're reading HN, why not come build software at Y Combinator itself! We are a small independent team of experienced engineers who try to make being funded by YC a great experience for founders. If the idea of designing software to help startups and the YC group partners sounds interesting to you, please reach out!

We work in Ruby on Rails, with a React/Typescript frontend. We move quickly and autonomously, and while we're not exactly doing rocket science (although we fund it), we're always experimenting with ways to make the YC batch better.

YC has excellent compensation and benefits (see more in the formal job description below). The team and the work life balance are great. About half of us are former founders and many of us are parents. And if you’re curious about startups (and possibly starting one someday), this job gives you amazing access to interact with YC’s programs, partners, and founders.

Please drop me a line if you're interested: casey@ycombinator.com

Details: https://www.ycombinator.com/careers?ashby_jid=00c6950f-341f-...

Work Authorization: This position does not support work authorization/visa sponsorship.

This is awesome! I built a similar open source version a few years ago for a friend with a fanless Air, but it foolishly used system notifications and a menu bar icon instead of sound:

https://github.com/caseymrm/notafan

This is great, I've done a few one-off go wrappers of objective C libraries [1][2] for an OSX menuapp framework I built [3].

This seems like a much more general and useful solution, excited to switch some things over to it!

[1] https://github.com/caseymrm/go-pmset

[2] https://github.com/caseymrm/go-smc

[3] https://github.com/caseymrm/menuet

I love the description of convolutions:

> I think of convolution as code reuse for neural networks. A typical fully-connected layer has no concept of space and time. By using convolutions, you’re telling the neural network it can reuse what it learned across certain dimensions.

The diagram is great too: https://attardi.org/pytorch-and-coreml#convolution

I believe this is what the author tried first in the post. He even links to this test UI where you can compare the "plain math" approach to the neural network:

https://attardi.org/pytorch-and-coreml-test-ui/

@steadicat I love this part:

> I think of convolution as code reuse for neural networks. A typical fully-connected layer has no concept of space and time. By using convolutions, you’re telling the neural network it can reuse what it learned across certain dimensions.

That's just where we put the ajax paths to keep a cleaner URL space

Interesting. However, it seems to block bots entirely: http://web.archive.org/web/*/http://www.quip.com

We just opened up the file at launch... Maybe caching or latency?

Evan is hilarious, check out all the copy on http://phabricator.org/ for an example.

However he should probably have been described as ex-Facebook, he left quite a while ago.

This made me crack up the first time I read it!

'Facebook engineers rave about Phabricator, describing it with glowing terms like "okay" and "mandatory".'

This made me crack up even more : "Written in PHP, so literally anyone can contribute, even if they have no idea how to program."

Evan is the man - he is without doubt one of the funniest people I knew at Facebook.

We're also big fans of Phabricator over here at MemSQL.

Thanks for the link. I may use Phabricator just so I have an excuse to dig through the rest of the docs. :) Well, that and it looks like it could be quite useful.

His flavor text is also a great read:

http://www.phabricator.com/docs/phabricator/#flavortext

Sorry I couldn't help more, internally a lot of us were surprised by this move also.

Nice. You need to linkify the tweets so users don't need to copy and paste the URLs to the actual job posts.

Agreed. Luckily Twitter makes it really easy: https://github.com/twitter/twitter-text-js

Thanks! This is even easier than asking on stack overflow!