Converting your O-1A into an EB-1A green card as a Machine Learning Engineers
The O-1A and EB-1A share vocabulary and much of their evidentiary architecture. The difference is durability: O-1A is a temporary status tied to a petitioner, EB-1A is a permanent self-petition. Most of your approved exhibits port over — with rework.
Built for: ML engineers, MLOps leads, and applied ML specialists who productionize models and build inference infrastructure.
Evidence portability: O-1A exhibits mapped to EB-1A criteria
Each row takes an exhibit you already filed and states what has to change before an EB-1A adjudicator will credit it.
O-1A evidence
Judging the work of others (review panels, hiring committees, competition juries)
EB-1A criterion
JudgingUpgrade action
Re-file the same invitations and completed review records, adding volume and selectivity evidence so the adjudicator sees sustained, not one-off, participation.
O-1A evidence
Critical or essential role for distinguished organisations
EB-1A criterion
Leading / critical roleUpgrade action
Convert employer support letters into outcome letters: name the mandate, the decision authority held, and the measurable result attributable to you.
O-1A evidence
High salary or remuneration relative to the field
EB-1A criterion
High remunerationUpgrade action
Benchmark total compensation against the correct OES occupation code and metro — $287,000 / $258,000 / $271,000 at the 90th percentile — with payroll and equity statements attached.
O-1A evidence
Published material about you in professional or major media
EB-1A criterion
Published materialUpgrade action
Add circulation or readership data and author credentials for each outlet; EB-1A adjudicators discount coverage without an evidenced audience.
O-1A evidence
Original contributions of major significance
EB-1A criterion
Original contributionsUpgrade action
Show downstream adoption — citations, deployments, standards references, or licensing — rather than the novelty of the contribution alone.
O-1A evidence
O-1A exhibit: Designing model-serving or training infrastructure used across the organization
EB-1A criterion
Original contributions of major significanceUpgrade action
Architecture docs, adoption metrics, cost/latency improvements, and independent letters on how the system influenced practice.
O-1A evidence
O-1A exhibit: Maintaining an open-source ML library, compiler, or serving framework
EB-1A criterion
Original contributions of major significanceUpgrade action
Download and dependency data, named production users, and expert letters explaining the technical novelty.
Satisfying the final merits determination
O-1A approval does not carry over. Even with three or more criteria satisfied, USCIS runs a separate final merits determination on the EB-1A petition and asks whether the record as a whole shows sustained acclaim and standing at the top of the field.
Sponsor freedom
No employer, no agent, no amended petition
An O-1A belongs to the petitioner. Change employers and someone has to file again; change the terms of your role materially and an amendment follows. Agent petitions add itineraries and third-party contracts to keep status intact.
EB-1A is filed by you. There is no job offer requirement, no labour certification, and no dependency on a sponsor's willingness to keep paying counsel — which matters most for Machine Learning Engineers who move between companies, consult, or found their own.
Related guides
How far does your O-1A record get you?
Run the free audit to see which approved exhibits already clear EB-1A criteria and which need rebuilding.