SWE-Marathon
[coming soon]
joan.so
I'm currently helping research labs on infra, long horizon, data, and RL.
[coming soon]
[coming soon]
Helping contributors getting tasks to the highest quality.
An universal standard for Benchmarking Models. ServiceNow.
[coming soon]
RL scaling on mathematical reasoning.
Co-founded:
Worked at:
I angel invest and invest via 3 different funds: a16z (scout), base10, and capitalize (YC-only). Mostly AI infra.
qlabs.sh - generalization research lab; daytona - agent sandboxes; stellon-labs - small speech models; trajectory - continual learning; bryel - lab automating harness engineering; cactus - on-device inference; lilac-ml - idle GPU market; piris labs - optical interconnect inference; polymath - RL environments; ishiki labs - teaching models proactivity; moss.dev - sub-10ms retrieval; squid - data center grid planning.
pave robotics - robotic crack sealing; vernerobotics - new skills in 30 samples; cortex - egocentric data collection; lightberry - human-robot interaction; hlabs - robotics in America; efference - depth-first cameras; origami robotics - manipulate-anything robots; human archive - human data collection; onerobot - world models for robotics; robodock - autonomous depots; voltair - autonomous EO drones; milliray - 20cm UAV detection.
phaselabs.bio - regenerating limbs; uncommon therapeutics - Rett Syndrome; inversion semiconductor - Moore's law; icarus - stratosphere airplanes; beyond reach labs - space infrastructure; terranox.ai - uranium discovery; kyten technologies - US battery packs.
blankbio - mRNA foundation models; strand ai - oncology foundation models.
dscribe - bulk inventory tracking; motives - consumer focus research; catchback cards - Pokemon cards.
and more ...
Reach Out