Neeraj Kumar
@neerajkumar.bsky.social
Chief Data Scientist @PNNLab. Writing about trustworthy agents, AI for science, physical AI, chips, compute and space infrastructure. Building a path to AGI.
The productivity gap may be a measurement problem before it is a capability problem. Biology needs fewer plausible outputs and more verified decisions per experiment.
August 6, 2026 at 11:52 PM
The productivity gap may be a measurement problem before it is a capability problem. Biology needs fewer plausible outputs and more verified decisions per experiment.
Prediction narrows the search space. Experiments tell us whether a structure works in a living system. The strongest AI biology loop will treat disagreement as data, not noise.
August 6, 2026 at 11:52 PM
Prediction narrows the search space. Experiments tell us whether a structure works in a living system. The strongest AI biology loop will treat disagreement as data, not noise.
Scientific foundation models will matter when they learn invariants that transfer across instruments and domains. Otherwise they risk becoming fast interpolators with excellent domain language.
August 6, 2026 at 11:51 PM
Scientific foundation models will matter when they learn invariants that transfer across instruments and domains. Otherwise they risk becoming fast interpolators with excellent domain language.
Congratulations team
November 22, 2024 at 10:35 PM
Congratulations team
I was in the line but then left as couldn’t stand anymore. Hope everyone make it to aquarium.
November 22, 2024 at 10:25 PM
I was in the line but then left as couldn’t stand anymore. Hope everyone make it to aquarium.