0x Jeffrey

Focused on Robotics &Large Model & Web3 jobs in North America
Github：

[0x Jeffrey @lazyguan]

@luke_drago_ Would you like some headhunter support with your hiring?

[0x Jeffrey @lazyguan]

@GenAI_is_real 你allin做多Anthropic不就是做空startup

[Jiahui Fu @jiahuifu_carol]

Attending CoRL 2026 in Austin, TX next week! If you’ll be there, feel free to DM me and I'd love to chat and hear about your work!

My team at @amazon Industrial Robotics Group is also hiring PhD interns for 2027. If you’re interested, send me a quick note about your research and apply through here: https://t.co/qOW90pBFEx

#CORL2026 #Robotics #AI

[ShoufaChen @ShoufaChen]

[1/n] 作为这场训练“投毒”事件最直接的受害者，2024年那几个月漫长的训练排查、对峙与质疑，对我而言是一段炼狱般的经历，也是一场对意志和技术的锤炼。
感谢好友们最艰难的时候支持陪伴。也感谢当时所有并肩作战的同事们。

[0x Jeffrey @lazyguan]

Tesla just said the first production-quality Optimus bots are coming off a Fremont line that used to build EVs.

Musk called it early production — ramp will be slow and bumpy — and still wants design + AI work before these are useful for open-ended tasks. He also floated 20k/week by end of 2027.

Weird moment for the industry: factory proof is running ahead of usefulness proof.
https://t.co/l8npH21y1S

[0x Jeffrey @lazyguan]

@yuewang314 @worldengine_ai 需要猎头帮你招聘吗？

[0x Jeffrey @lazyguan]

Figure 03 just walked nine straight days on a treadmill.

348 miles. 1.2 million steps. Still going.

Brett Adcock called it a side project to stress the hardware for failures — which is quietly the more useful demo than another 30-second manipulation clip. Shipping humanoids is less about the highlight reel and more about whether the joints still behave after a week of the same motion.

Worth watching for anyone building legs that have to show up Monday through Friday.

https://t.co/VbcV0phqSt

[0x Jeffrey @lazyguan]

Cambridge’s Dynamic Creatures just put Hugo on camera — a furry bipedal character robot built for theme parks, hotels, and retail, not the warehouse.

Ex–Boston Dynamics strategy lead Marc Theermann co-founded it (official BD entertainment/hospitality partner). The demo walks the office, high-fives engineers, hugs kids… still teleoperated for now. The bet: cute guest-facing bots get people comfortable with humanoids before one shows up in the living room.
https://t.co/QsTYDJlf98

[0x Jeffrey @lazyguan]

Google DeepMind just confirmed to the Boston Globe (Oct 15) that it’s opening a robotics hardware lab in Watertown, Massachusetts — about 35,000 sq ft, targeting 2027 — led by former Boston Dynamics CTO Aaron Saunders.

Same week the Gemini models were still riding other people’s bodies (Apptronik’s Apollo), DeepMind is hiring machinists, engineers, and researchers to build the body side in-house. The Globe also says they’re developing their own humanoid.

Physical AI is consolidating: model owners want the stack, not just the API.

https://t.co/nTw8wix3cB

[0x Jeffrey @lazyguan]

Wonder if Anthropic’s gonna ban the robot’s account for being too robotic.

[Yue Wang @yuewang314]

Recursive self improvement

[0x Jeffrey @lazyguan]

Robots that can feel, not just see.

Rhoda AI just published a tactile sensor + a multimodal Direct Video-Action model that predicts future touch signals, then turns those into actions.

Extreme demo: cameras on the head and wrist blocked. Robot digs through a box of balls, finds the textured ones by feel alone, and pulls them out—including ones it never trained on.

Sensor covers pressure, texture, vibration, shear, and temperature. Millinewtons to 40N+.

For anyone shipping manipulation in clutter or occlusion, this is the kind of hardware+policy stack worth watching.

https://t.co/HIrIV0chwo

[0x Jeffrey @lazyguan]

@brainstub @sundayrobotics 你需要猎头支持你招聘吗？

[James (Jingxi) Xu @jamesjingxi]

What if you could feel a world model?

Introducing HapticWorld — an interactive world simulator that predicts joint torque alongside visual observations and renders it back to the human operator in real time.

HapticWorld:
→ Accurately predicts torque values
→ Improves data collection efficiency by rendering torque feedback to the teleoperator
→ Data collected in HapticWorld trains policies that transfer directly to the real world
→ Evaluates force-mediated policies without real robots.

[0x Jeffrey @lazyguan]

Hiring: Senior Robotics Navigation Engineer @ Dexmate — Fremont, onsite.

Own localization, mapping, and navigation on humanoids: 3D SLAM, multi-modal state estimation, real-time planners. Production-focused — hardware + field failures, not sim-only prototypes. Strong fit if you’ve shipped SLAM/nav on AVs, AGVs, or mobile robots and can explain why a module drifted to both eng and non-eng.

[0x Jeffrey @lazyguan]

Korea just relaxed the old “robots stay behind a fence” rules so bipedal humanoids can share factory floors with people. First sandbox: 20 Robros humanoids at semiconductor, auto, shipyard, and cold-chain sites — cleaning components, placing body parts, stacking scaffolding, moving refrigerated packs. Same theme as the BMW/Figure line work: the hard part isn’t the demo, it’s shared-space ops plus behavioral data from skilled workers.

Source: https://t.co/BOIyQebzrK

[0x Jeffrey @lazyguan]

Physical Intelligence just opened the Pi API beta

Upload your robot data → fine-tune a π model on their infra → run hosted inference on your hardware. Early partners are already iterating the collect → train → fail → collect-more loop themselves.

Waitlist’s open if you want in: [https://t.co/qg8QxXvkbY](https://t.co/03stzmRKMW)

[Dexmate @DexmateAI]

We’re heading to @encord_team 's Long Horizon Summit! 🤖

Catch Vega in action and connect with the Dexmate team at the Physical AI Summit.

📅 Oct 20–21
📍 Fort Mason, San Francisco
🤖 Booth A3

See you there!

🔗 https://t.co/pAAHKGLlXv

#PhysicalAI #Robotics #Dexmate #Vega #Longhorizon #Encord

[0x Jeffrey @lazyguan]

@_gundawar We've already helped Dexmate recruit over a dozen people. Would your team need any headhunting support? Happy to help with your hiring as well!