Nathaniel Saul

once topology. now robopsychology.

[Nathaniel Saul @NathanielSaul]

Pretty much all my projects recently

[New Scientist @newscientist]

🖊 by @tomgauld

[Nathaniel Saul @NathanielSaul]

@EvanDeKim @ctralie Rad! Totally into it. Let me know how I can help,

[Nathaniel Saul @NathanielSaul]

Interesting that most PMs that responded indicate they wish they’d taken more CS courses while the narrative for SWEs is that courses aren’t needed.

[Nathaniel Saul @NathanielSaul]

@sarahcat21 If we use construction as an analogy, the library is the building material and tools, the framework is the architectural blueprints, and the platform is the city, it’s building codes, and regulations.
All three have to work well together to ensure building last long enough.

[Nathaniel Saul @NathanielSaul]

Bummed that I don’t have any neurips acceptances this year. Maybe next year I’ll actually submit something.

[🐝 @bullhornbitch]

the grand ronde tribe reclaimed willamette falls, and they’re tearing down the old paper mill to build a community center!!! what a win 👏🏽!!

[Nathaniel Saul @NathanielSaul]

@vboykis Python, SQL, and does HCL count?

[Arvind Narayanan @random_walker]

When a machine learning system uses argmax to select outputs from a probability distribution — and most of them do — it's a clue that it might be biased. That's because argmax selects the "most probable" output, which may amplify tiny data biases into perfectly biased outputs.

[Nathaniel Saul @NathanielSaul]

@teej_m Ah jeeze can you please create a dbt package for this so I can use it without having the chills from writing it out myself.

[Nathaniel Saul @NathanielSaul]

@patrickCDEmath @leland_mcinnes controversial in academia maybe.. TDA is exciting for mathematicians, but I don’t think anyone has successfully incorporated it into product. Except maybe umap?

[Nathaniel Saul @NathanielSaul]

This was a beautiful story

[Nathaniel Saul @NathanielSaul]

@karlhigley I’d posit that the particulars about side information doesn’t transfer as well between applications, so there’s not much benefit in learning from others. Everyone can have the same model but different side info.

[Nathaniel Saul @NathanielSaul]

@johncutlefish How do you balance short term thinking and the ideal current state thinking you’ve promoted in other tweets?

[Nathaniel Saul @NathanielSaul]

If you’re interested in TDA, we’re always looking for help with @scikit_tda! There are lots of interesting things still to implement and Chris and I try to be helpful with new contributions.

[Dr Finished and Tired @DoneWithMaths]

Hey all, I’m a pure math PhD in homotopy theory and homological algebra, and I want to get a job one day so I need to learn to code. Anyone have any advice on projects which my pure math addled brain won’t find impossible?

[Dr Finished and Tired @DoneWithMaths]

Hey all, I’m a pure math PhD in homotopy theory and homological algebra, and I want to get a job one day so I need to learn to code. Anyone have any advice on projects which my pure math addled brain won’t find impossible?

[Nathaniel Saul @NathanielSaul]

@Aras121 @scikit_tda @ctralie Yeah there is an example of running a sublevel set filtration on an image in the https://t.co/rITID4tESf documentation site!

[Nathaniel Saul @NathanielSaul]

@rjurney @rit But it is a singular forum. There are so many more. So many that trying to reduce OSS engagement to just at SO votes is ridiculous. Sure it’s a metric, but I think it’s a terrible one.

[Nathaniel Saul @NathanielSaul]

@rjurney @rit What? By that same logic, the documentation does my job. SO answers FAQs, it does not build applications.

[Nathaniel Saul @NathanielSaul]

@rjurney @rit It’s like trying to say something about someone’s contribution to community development based on their likes on Facebook.

[Nathaniel Saul @NathanielSaul]

@rjurney @rit How does a vote relate? Sure, it’s easy to vote, but is just one tiny signal. OSS is so much more than that and I think it’s pretty poor proxy.