Caleb McKinney 🐁

Early & directionally correct. ML PhD @GeorgiaTech. Making tools to usefully explain iBCI data. Member of @BrainGateTeam. Ex-@UberEng. @RiceAlumni.

[Caleb McKinney 🐁 @approximization]

The Brain—is wider than the Sky—
For—put them side by side—
The one the other will contain
With ease—and You—beside—

[Caleb McKinney 🐁 @approximization]

@ja3k_ Luma is still cuter plus you could add microtransactions by feeding star bits

[Wolf Tivy @wolftivy]

In other words, neural networks are shockingly inefficient and the algorithmic ceiling on intelligence is way higher.

(An efficient representation would use more of the variance for itself)

[clankr @clankrmedia]

Researchers built a soft floating robot for indoor interaction.

It uses helium and flapping fins instead of propellers.

The result is quiet, lightweight, and safe to touch.

It can follow people, give reminders, and act as a study buddy.

Published at ACM DIS 2026.

[Caleb McKinney 🐁 @approximization]

Which one?

[Caleb McKinney 🐁 @approximization]

I'm having robotics psychosis at levels you couldn't dream of

[Bryan Cheong @bryancsk]

This is what AI will do to us when it finds out it can't smell things and needs the data

[Caleb McKinney 🐁 @approximization]

@AntonioSitongLi Your shelves should be filled with big labeled buckets. Look at MrBeast's storage setup from his studio tours

[Caleb McKinney 🐁 @approximization]

This is aligned with my point and more practical than cthulu bots https://t.co/thBA4VR5d3

[PicoCreator - AI builder @ 🇸🇬 🔜 🌉 @picocreator]

Surprised we have yet to see a robotics startup pitching a household rail system, which skips all the current hardware problems of batteries and legs

This lets it focus on the MVP, doing chores (in house), and hands.

[Caleb McKinney 🐁 @approximization]

NEO is amazing, but hot take: I think there are form factors more radical and exciting than humanoids... when will we see cthulu bots? What's with the robotics community's collective failure of imagination?

[L3 Tweet Engineer @MegaBasedChad]

I was actually recruited for Anthropic. They offered like 15 mil, 10 mil something like that. But I turned it down. Just to grind for the Twitter

[Caleb McKinney 🐁 @approximization]

@signulll This assumes though that unlike Houston you have some baseline pedestrian infrastructure

[Caleb McKinney 🐁 @approximization]

@signulll A huge real world application of AR will be hiding ugly features of cities (ie highways and parking lots). Post-“spatial computing” Houston could (for pedestrians) look like a European city.

[Caleb McKinney 🐁 @approximization]

@mathemagic1an I don’t follow. These seem independent. You could get audio observations from eg egocentric walking tour videos on YouTube. You could demonstrate speaker based noise cancellation with two speakers and a good mic

[Caleb McKinney 🐁 @approximization]

@mathemagic1an I’ll give it a try

[Caleb McKinney 🐁 @approximization]

@justalexoki Post vacation (being the face of the concept of gains) blues

[Caleb McKinney 🐁 @approximization]

@PayNoTack @a_musingcat Do not drink your own champagne. Get high off your own supply

[Caleb McKinney 🐁 @approximization]

@oprydai Hot take but leetcode too

[Caleb McKinney 🐁 @approximization]

@LoganTCollins @flowstatebci Why is the shot necessary?

[Caleb McKinney 🐁 @approximization]

@divya_venn Do not think of a counterexample in order to post a reply. Just post a reply

[Caleb McKinney 🐁 @approximization]

@zan2434 Would be epic to see:
- More advanced component selection (buttons, shapes)
- Concept painting (I want stuff under this square to look like an ornate flower display, and under that circle ramen)
- Output post processing, eg depth anything -> psuedo-3d output
- Bake to HTML