Mechanize
@MechanizeWork
We're a software company building RL environments to power the full automation of the economy
We're hiring software engineers. $500k base. x.com/i/jobs/1919892…
Automating math would generate less than 1% as much value as automating software engineering. Perhaps AI labs should focus less on chasing gold medals and focus more on the hard problem of automating SWE.
Many software engineers want to move into AI but think they need to learn ML first. We are offering an alternative: researcher-level pay but without any need for prior ML experience. We're seeking highly talented engineers with traditional SWE experience. x.com/i/jobs/1919892…
The bitter lesson of the past decades is that the best algorithms are discovered by applying massive compute for search & learning. Progress in architectures has been incremental, while data-driven innovations have been highly impactful. We expect this trend to continue.
the architecture isn't the bottleneck, it's the data
xAI is probably the first to spend as much compute on RL as on pretraining. The easy gains from shifting compute to RL are now gone. With this arbitrage closed RL scaling will slow. Progress will now come from the quality and realism of RL environments rather than mere scaling.
