Framing the proposed endeavor
Generic endeavors fail. 'Continue working as a software engineer' does not map to Dhanasar Prong 1. Successful endeavors are scoped to a specific technology area with verifiable national-interest implications.
Strong examples: 'Advance the security of U.S. critical-infrastructure software supply chains by developing and operationalizing reproducible-build tooling at scale,' or 'Improve the reliability of large-scale distributed AI training systems used by U.S. semiconductor and cloud providers.'
Evidence categories that work
Software engineers commonly assemble: GitHub project metrics (stars, downstream dependents, downloads via npm/PyPI), conference talks at major venues (KubeCon, NeurIPS, USENIX, ACM), patents, internal company impact metrics, and recommendation letters from independent senior engineers.
- Open-source adoption metrics (downloads, downstream dependents, GitHub stars and forks)
- Talks at top-tier conferences (Strange Loop, USENIX, ACM, IEEE, KubeCon, NeurIPS)
- Granted patents in the relevant field
- Documented production impact at scale (users served, revenue, latency improvements)
- Letters from independent engineers at peer companies and labs
- Citations of technical publications or whitepapers
Prong 3 arguments for software engineers
PERM labor market tests do not capture the value of a software engineer with a specific track record on a specific endeavor. A general 'senior software engineer' job posting cannot identify the petitioner. Frame Prong 3 around the wasteful or impractical nature of trying to test the labor market for the petitioner's actual contribution.