Kai Chan

Lead Technical Program Manager, Data Engine @ Sunday | Ex-Amazon

[Kai Chan @kaichan415]

A few hours of post training data.

That was enough. Memo picked up a new skill because recovery, generalization, and robustness were already there.

Pretraining did the heavy lifting.

When your base data is diverse enough, post training is not the foundation. It's where all that upstream work pays off. ACT-2 is proof of that.

[Tony Zhao @tonyzzhao]

Introducing ACT-2 Preview

The first robotics model to unify broad generalization with high reliability. A single fine-tuning example can teach Memo a new behavior that generalizes.

Zero shot, real unseen homes, 99% success rate.

[Kai Chan @kaichan415]

@henry_yu_01 @sundayrobotics Thank you for all you do Henry!!!

[Kai Chan @kaichan415]

Your next adventure might involve robots. We’re hiring 🤖💪

[Perry Jia @perryzjia]

My Data Operations team is hiring for a bunch of roles @sundayrobotics! Join us in building general purpose robots! 💪

Data Annotation Lead (leadership level role): Own the operation side of data annotation and scale the team behind it
https://t.co/T2Xt8wKo3c

Technical Program Manager, Data Engine: Own the project side of either data annotation or data collection
https://t.co/eHFbiyoCgG

Eval Ops Program Manager: Build the operational system for robotics evals
https://t.co/N9338UA4Dq

Strategic Projects Lead: Build undefined problems from 0 -> 1
https://t.co/nUyPR8dDVh

Triage Associate: Serve as the first line of defense in debugging hardware and software issues across our robot fleet
https://t.co/cbj53z8bKc

[LeRobot @LeRobotHF]

Hey @sundayrobotics, we also tried that! But ACT2 seems to work way better, congrats on the release!

[Kai Chan @kaichan415]

New life experience unlocked 🔓 Sprinting into a restaurant carrying your nephew who “didn’t have to pee” five minutes ago, convinced you’re going to make it, then suddenly feeling warm pee drip down your shorts while the waiter tries not to laugh 😭

[Kai Chan @kaichan415]

1,000+ Memory Developers don’t make ACT-2 robust despite their differences.

They make it robust because of them.

Turning that diversity into consistently high quality data is hard and incredibly special!

Huge respect to @perryzjia and every MD who helped scale this from Craigslist to production. Proud to lead this part of the mission.

[Perry Jia @perryzjia]

1/N Two years ago, we recruited our first Memory Developer off Craigslist and onboarded them in a public library.

Today, more than 1,000 Memory Developers have helped us build the data engine behind ACT-2.

Here's how we got from that library to here 🧵

[Cheng Chi @chichengcc]

We just hit a weird milestone: our model became more reliable than your average home WiFi.

Just like everybody else, we thought cloud inference was the obvious choice. Yet 2 days into the ACT-2 eval, our mind completely changed.

If our hero @ArpitKalla didn’t cook, this video wouldn’t exist 🧵

[Shuran Song @SongShuran]

One year ago, I wouldn't have imagined that laundry folding would be the first "solved" task for home robots. The world is moving soooooo fast!

[Kai Chan @kaichan415]

#cakechain

[S G @StefAtSunday]

Generalization is a piece of cake when your pretraining data is diverse af. Shout to our 🐐 TPM @kaichan415 🦾🍰✨️

[Kai Chan @kaichan415]

@nishantsdesai Turns out letting the data breathe was the unlock. Loving this ACT-2 era 🔥"

[Kai Chan @kaichan415]

@perryzjia @sundayrobotics Happiness at scale = magic. Proud to be part of this 🚀 !!!

[Kai Chan @kaichan415]

@alpercanbe Memo just leveled up laundry forever!

[Kai Chan @kaichan415]

@JunyaoShi @JunyaoShi loved working so closely with you on ACT-2.

Going from hundreds of demos that broke instantly to one example plus five minutes of fine tuning that actually generalizes is wild!

[Sunday @sundayrobotics]

We’re hiring in-house Memory Developers to shape how Memo perceives and interacts with the world.

The Role:
📍 On-site & flexible (3–5 days/week).
👟 Record high-quality task demos (e.g. folding t-shirts, arranging shoes, etc).
🧠 Beta-test new hardware and software.
📈 High growth potential into management.

Apply below if you’re detail-oriented, curious about AI, and interested in contributing to the future of autonomous home robotics.

[Kai Chan @kaichan415]

@keval_shah01 So excited for this moment 🚀

[Kai Chan @kaichan415]

@chichengcc So excited for this moment 🚀