Tinygrad says it is ending a bounty program that once helped identify potential hires because AI-generated submissions have made the results less informative. The company describes a broader hiring problem in which automated resume spam and AI-assisted pull requests can waste the time of both applicants and screeners, including submissions from people who barely understand the task. Instead, candidates are asked to contribute openly to tinygrad over time and demonstrate that they understand the project’s goals. The company says it welcomes AI in software-development workflows, but AI use alone will not make it hire someone it would previously have rejected, or reject someone it would previously have hired. Tinygrad says it values deletion, clearer design, deep bug fixes, and regression tests over simply adding features. Candidates are encouraged to read the project’s Discord discussions, commit history, and meeting archives before contributing. The company presents tinygrad as a long-term effort to challenge technology companies’ infrastructure rents by commoditizing computing capacity. The role is remote or hybrid, with offices in San Diego and Hong Kong, a weekly one-hour meeting, and a biweekly three-sentence sprint summary. Full-time compensation is listed as $75,000-$150,000 plus 0.1%-0.5% equity, and most work is open source and public.
