Stephen James

@stepjamUK

CEO @Neuracore_AI | Assistant Professor @imperialcollege | ex-Director of Dyson Robot Learning Lab | Postdoc @UCBerkeley w/ @pabbeel | PhD ICL w/ @ajdDavison

London, England

stepjam.github.io

Joined January 2010

Pinned
Stephen James
@stepjamUK
Sep 24, 2025
𝗔𝗳𝘁𝗲𝗿 𝟭𝟬+ 𝘆𝗲𝗮𝗿𝘀 𝗶𝗻 𝗿𝗼𝗯𝗼𝘁 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴, from my PhD at Imperial to Berkeley to building the Dyson Robot Learning Lab, one frustration kept hitting me: 𝗪𝗵𝘆 𝗱𝗼 𝗜 𝗵𝗮𝘃𝗲 𝘁𝗼 𝗿𝗲𝗯𝘂𝗶𝗹𝗱 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗶𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝗼𝘃𝗲𝗿 𝗮𝗻𝗱
00:00
Stephen James
@stepjamUK
Aug 13
Excited to be hosting our first event at @Neuracore_AI HQ next week for Automate Further! If you're working in robotics, factory automation, or looking to get more out of your production lines, this is your last chance to sign up and join us. luma.com/5a584dfd
Neuracore
@Neuracore_AI
Aug 13
There's a gap between UK robotics research and the factory floor. On 18 August, we're closing it. Sarthak Das from Neuracore breaks down what to expect at Automate Further, where FANUC, @rarukautomation, @HALrobotics, @ARIA_research, @theautomateuk and @the_MTC_org will come
00:00
Stephen James
@stepjamUK
Aug 13
There's a nice failure mode in inference-time steering for VLAs that most methods just don't handle: they apply the same intervention at every timestep, whether or not the policy actually needs help. Generate more candidate actions, rephrase the instruction, sample from a wider
00:00
Stephen James
@stepjamUK
Aug 11
More data and a bigger model should make a robot policy better. It does, up to a point, and the interesting part is where that point sits. Most robot learning research only answers half of that question. You get pre-training scaling curves with no evidence the gains survive
00:00
Stephen James
@stepjamUK
Aug 11
One week until #AutomateFurther: luma.com/5a584dfd I'll be speaking about the shift from hand-coded control to policies learned via imitation, and what it actually takes to get that from the lab to reliable deployment on a factory floor. That gap is the problem Neuracore