For data scientists and ML engineers

O-1, EB-1A and NIW for data scientists and AI/ML engineers

Papers, conference reviewing, open source and a key role on a model team often map to the criteria. Here is which route fits and what evidence usually counts.

Last checked October 2026 · Sources

PhD needed for EB-1A
No
O-1A legal fee from
$6,999 flat
EB-1A legal fee from
$8,995 flat
EB-2 NIW legal fee from
$7,995 flat

Quick answer

Which visa or green card fits a data scientist or ML engineer?

Data scientists and AI/ML engineers usually compare O-1A, EB-1A, EB-2 NIW and H-1B. EB-1A has no degree requirement in 8 CFR 204.5(h). You need a major award or 3 of 10 criteria, then a review of your whole record. EB-1A and NIW let you file for yourself.

  • Conference papers and peer review for conferences are common ML evidence.
  • O-1A skips the H-1B lottery, but an employer or agent must file.
  • NIW needs an advanced degree or exceptional ability first.
  • A job title in AI does not qualify you on its own.

Compare routes

O-1A vs EB-1A vs NIW vs H-1B for data and ML

The fit column is practical guidance. A partner-firm attorney confirms what fits your case.

RouteWhat you getFit for data and MLKey hurdleOur legal fee from
O-1AWork visa, up to 3 years at firstStrong record, need to work soon, no lotteryA major award or 3 of 8 criteria, plus an advisory opinion. An employer or agent files.$6,999
EB-1AGreen card, no sponsorCited papers, adopted tools, a critical role, high pay3 of 10 criteria, then a final merits review for sustained acclaim$8,995
EB-2 NIWGreen card, no sponsorMaster’s or PhD, or bachelor’s plus 5 years of progressive experience, with work of national importanceThe 3-part test in Matter of Dhanasar, 26 I&N Dec. 884 (AAO 2016)$7,995
H-1BWork visaEmployer will sponsor youCap-subject jobs go through a yearly lottery weighted by wage level, starting FY 2027Quoted on request

Advanced degree definition: 8 CFR 204.5(k)(2). O-1A criteria: 8 CFR 214.2(o)(3)(iii). Premium processing (Form I-907) means USCIS acts within 15 business days for O-1 and EB-1A, and 45 for NIW (USCIS). Government fees are separate. See pricing.

Your evidence

Typical data and ML evidence, by criterion

Practical guidance on where data and ML work usually fits. What counts in your case is decided on your documents.

CriterionTypical data or ML evidenceWhat it must show
Scholarly articlesPapers at ML conferences or in journals, technical articles in major trade publicationsYou are a listed author, writing for experts in your field
JudgingReviewer or area chair for ML conferences and workshops, Kaggle or hackathon judgingYou were asked and you completed the review
Original contributionsOpen-source libraries others adopt, models or methods others build on, patents others useMajor significance to the field. A paper or patent alone may show originality, not impact.
Leading or critical roleOwning a key model, data platform or ML teamYour performance in the role, not your title, and the organization’s reputation
High salaryPay records, equity, offer lettersHigh pay compared with others in your field and location
AwardsBest paper awards, competition winsNational or international recognition, and you are a named recipient

Sources: 8 CFR 204.5(h)(3) and the USCIS Policy Manual, Vol. 6, Part F, Ch. 2. The manual says the person’s “performance in the role,” not the title, decides whether a role is critical. Read more in EB-1A for data scientists and ML and citation record evidence.

Get started

Where data and ML people usually start

  1. Audit your record

    List papers, citations, review invitations, talks, awards, open-source adoption and pay.

  2. Pick 2 or 3 criteria you control

    Conference reviewing, critical role and high salary are often within reach. See which criteria to build first.

  3. Collect outside proof of impact

    Citations, downloads, adoption by unrelated companies and letters from people outside your team carry more weight than your own description.

  4. Choose the route with an attorney

    Practical guidance: O-1A now and a green card later is a common plan. A partner-firm attorney decides what fits your case.

FAQ

Questions people ask

Yes. 8 CFR 204.5(h) has no degree requirement. You need a major award or 3 of 10 criteria, and then USCIS weighs your whole record.

Peer review for a conference in your field can count as judging the work of others. Keep the invitation and proof you completed the reviews.

No set number exists. Citations are one way to show your work had impact. Practical guidance: adoption in products or open source can also help.

It can meet one criterion if your pay is high compared with others in your field and location. You still need at least two more criteria for EB-1A, and then the review of your whole record.

If you have an advanced degree or exceptional ability and your work meets the Dhanasar test. Read NIW for STEM and entrepreneurs.

Yes. EB-1A has no employer requirement, so you can file for yourself while you work. Ask a partner-firm attorney how it fits your current status.

See where your data or ML record stands

Upload your resume and see your standing on each criterion. No account or card. A partner-firm attorney confirms eligibility before anything is filed.

Start free evaluation

Your first step

Check your record against every criterion

Start free evaluation

Free. No account, no card. Not legal advice.