Jayesh Singla

Exploring the realm of ML and robotics | MSCS @ CMU | IIT Bombay CSE

[Jayesh Singla @SinglaJayesh]

Robots are leveling up!

With the capability to learn from one demonstration, S1 is nearing superhuman territory (I would personally forget half the steps if I watched a 10-minute long video). Using S1, time to deployment is going down from weeks to mere hours!!

S1 is not just memorizing the video, it understands every step and adjusts to differences in the real world - just as a human would.

As training scales up, I can’t wait to see how far we can push the boundaries of what’s possible...

[Skild AI @SkildAI]

Introducing S1, our new foundation model that learns from one example.

It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning.

Watch S1 operate in real-time via in-context learning:

[Jayesh Singla @SinglaJayesh]

Really neat idea to preserve the generality of representations in VLAs. Seems like we were really under-utilising the capabilities of VLMs while training VLAs.

Impressive gains with such low overhead tweaks!

[Unnat Jain @unnatjain2010]

"The generalization you want in your VLA has to be in the expert demos!" folks say.
"Didn't we train VLM backbones on trillions of tokens already for it?", we ask.

We set out to reclaim some of this (free) OOD generalization in VLAs. Just need to think beyond BC! 🥲

https://t.co/nBjmexOzXM
Two edits on top of BC. Same demos. No extra data.

[Unnat Jain @unnatjain2010]

"The generalization you want in your VLA has to be in the expert demos!" folks say.
"Didn't we train VLM backbones on trillions of tokens already for it?", we ask.

We set out to reclaim some of this (free) OOD generalization in VLAs. Just need to think beyond BC! 🥲

https://t.co/nBjmexOzXM
Two edits on top of BC. Same demos. No extra data.

[Skild AI @SkildAI]

Introducing S1, our new foundation model that learns from one example.

It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning.

Watch S1 operate in real-time via in-context learning:

[Akash Sharma @akashshrm02]

Presenting PTLD: an approach to learning tactile dexterous policies without ever simulating the tactile sensor.

Tactile is essential for performing highly dexterous manipulation. However, collecting tactile observations reliably has been a fundamental bottleneck: (1) teleoperating a multi fingered hand for dynamic tasks is challenging, making sim-to-real imperative, (2) one can’t realistically simulate tactile today

[Skild AI @SkildAI]

Exciting news! 🎉 Our CEO & Co-founder, Deepak Pathak (@deepakpathak), received the PAMI Young Researcher Award at #CVPR2026 this week.

Among the highest honors in computer vision for early-career researchers, the award recognizes groundbreaking contributions that have a lasting impact on the field of AI.

Congratulations, Deepak!

[Skild AI @SkildAI]

We have acquired Zebra Technologies’ robotics arm (formerly Fetch Robotics).

This is what happens when orchestration meets intelligence -- a major step toward fully autonomous warehouses.

More robots. More environments. One unified brain.

[Unnat Jain @unnatjain2010]

CRAFT hand🫳
1. Achieves all 33/33 dexterous grasps > 2x-20x $$ hands!
2. < $600
3. Handles fragile objects
4. Durable under contact
5. Open-sourced https://t.co/lCBAeklcVn

@leo_lin6 & @shivanshpatel35 (on market; hire him🚀) will happily share anything else that you may need. Details in 🧵

🌟 Big shout out to @kenny__shaw (Leap & v2), @irmakkguzey (RUKA), @orcahand (ORCA), and many others who helped build this open research community. Thank you!

[Gauri Gupta @gauri__gupta]

We @neosigmaai @RitvikKapila are building the future of self-improving AI systems! By closing the feedback loop between production data and system improvements, we help teams capture failures, convert them into structured evaluation signals, and use them to drive continuous improvements in agent behavior.

We show how our system works on Tau3 bench across retail, telecom, and airline domains. Agent performance on the validation set (with a fixed underlying model, GPT5.4) improves from 0.56 → 0.78 (~40% jump in accuracy).

[Skild AI @SkildAI]

Nearly every system today, from energy to chips to food, is bottlenecked by scarce human capital.

We are changing that by building AI-powered industries of the future.

Check out Skild Brain robustly assembling GPU racks, a highly precise task, live at #NvidiaGTC.

[Skild AI @SkildAI]

Robotics is a data problem.

Today, we’re partnering with @ABBRobotics, @Universal_Robot, and @NVIDIARobotics to deploy the Skild Brain across real-world industries from manufacturing to factory lines.

This will help us build the world’s biggest data flywheel for physical AI.

[Skild AI @SkildAI]

India, we’re here! 🇮🇳🇺🇸

From one Silicon Valley to another, we’re thrilled to open our newest office in Bengaluru.

[Skild AI @SkildAI]

Let robots clean in peace 🧺🤖✨

[Jayesh Singla @SinglaJayesh]

Reasoning about tasks, intent, objects, distractions, perturbations—just by watching videos.
When perception turns into understanding, you know AGI is getting closer....

[Skild AI @SkildAI]

Humans learn by watching. Robots should too.

[Skild AI @SkildAI]

Humans learn by watching. Robots should too.

[Skild AI @SkildAI]

Happy Holidays!! 🎄🎄

[Skild AI @SkildAI]

giving our humans (and robots) a break at skildsgiving ‘25

[Jayesh Singla @SinglaJayesh]

@SkildAI

[Skild AI @SkildAI]

can't catch me now 🏃💨🪚🪚

[Skild AI @SkildAI]

We built a robot brain that nothing can stop.

Shattered limbs? Jammed motors? If the bot can move, the Brain will move it— even if it’s an entirely new robot body.

Meet the omni-bodied Skild Brain: