ML engineer: US talent visa (O-1, EB-1A)
The "ML engineer" profession in the "IT / software" field: whether the talent visa fits, which criteria are realistically achievable, what to collect and what AAO decisions in the field show.
Updated: Author: Alina Kanametova"IT / software" field overview
Key answer
Does the talent visa fit for an ML engineer?
Yes, an ML engineer is one of the strongest IT profiles for the talent visa. The usual criteria are original contribution (models, open source, AI infrastructure used by others), high remuneration (typical for ML/AI roles when benchmarked against the market), judging (paper review, hackathons, grants) and a leading role. A title alone is not a criterion - you need documented impact beyond routine engineering work.
Usually built on:
Which route fits for an ML engineer
| Visa | Who it fits | Next step |
|---|---|---|
| O-1 | Who it fits:Original contribution, High remuneration, Judging, Leading role: through a US employer or agent, no green card | Find your visa |
| EB-1A | Who it fits:Original contribution, Judging, Leading role: a green card by self-petition | Estimate your odds |
| EB-2 NIW | Who it fits:Original contribution, Judging: framed as a project of national importance (Dhanasar) | Build the packet |
Achievable criteria in "IT / software"
- Original contributionrealistic for IT if the contribution went beyond ordinary engineering work and had a noticeable impact on the product, industry, open source, security, AI, data infrastructure, or business processes.
- Leading roleachievable, but you need to prove influence on a key project and the reputation of the organization or product.
- Judginga realistic criterion for IT through evaluating competitions, hackathons, grants, accelerators, technical articles, open source projects, or candidates in professional programs.
- Membershipspossible, but it works only if admission requires selection based on achievements, experience, recommendations, or expert status, not just payment of dues.
- Media about youachievable if there are materials in independent industry sources showing the significance of the development, product, research, or expertise.
- High remunerationoften applicable in IT, especially for senior, staff, principal, architect, ML/AI, security, and leadership roles, if there is market comparison and supporting documentation.
What to collect: evidence for an ML engineer
- For contribution: independent expert letters and metrics showing your model or AI infrastructure was used by other teams, companies or the open-source community.
- For judging: invitations to review papers and competitions, confirmations of judging at hackathons, grants or accelerators, organizer letters.
- For remuneration: contracts and statements comparing your compensation with market data for ML/AI roles.
- For contribution: independent expert letters showing that the development was used by other teams, companies, the open source community, or affected product metrics, security, cost, speed, or scalability.
- For a critical role: letters from managers and outside specialists, documents about a key project, architectural decisions, ownership, impact on revenue, users, SLA, compliance, implementation of AI/ML, or infrastructure.
- For judging: invitations to serve on a jury, confirmations of participation in evaluating work, judge selection criteria, lists of reviewed projects, certificates, letters from contest organizers or professional programs.
- For membership: association rules that specify applicant requirements, confirmation of membership level, recommendations, proof of selection, and an explanation of why the membership reflects recognition in the profession.
- For publications: articles, interviews, product reviews, mentions of an open source tool, professional podcasts, or industry materials where the specialist or their work is the subject of discussion, not an ad insertion.
- For high pay: employment agreements, offer letters, equity/bonus documents, tax records, payroll, income statements, and market reports for comparable IT roles, region, and seniority level.
Associations
- from real cases
- from real cases
- from real cases
- from real cases
- from real cases
- from real cases
- from real cases
invite-only, fee-based - weaker as a membership criterion
- from real cases
IAITP (International Association of IT Professionals)
- from real cases
IAHD / IAHQD (International Association of Honored Developers)
- from real cases
International Association of Highly Qualified Software Developers
- from real cases
RAEC / AEC (Russian Association for Electronic Communications)
- from real cases
IT Ukraine Association
- from real cases
Organization of Georgian IT Specialists
- from real cases
AITEX
- from real cases
National Association of Digital Economy
Mind the Product
product-management community
Products That Count
product-management association
Judging
Conference / journal peer review (IEEE, ACM)
Where people in "IT / software" publish, win and belong
Most often missing: Original contribution
Community questions
- Foreign-company salary: which country do I compare against?You compare your income against the market of the country where you actually worked and paid taxes, for your occupation. If a foreign company pays you remotely, the community advises comparing against the local market for the role, backed by several statistics sources. Officers usually dispute the comparison method, not the number.
- What documents prove the high-salary criterion?The working combo is proof of your income (contracts, bank statements, tax records) plus a market comparison from several independent sources - official statistics, job aggregators, analytical reports. The community advises showing not just the average but the top of the salary range for your occupation and region.
- How do freelancers close the high-remuneration criterion?Being self-employed is not a blocker: official income documents work - tax filings, bank statements, client contracts. The harder step is the market comparison. The community advises comparing your personal income against salaried professionals in the same occupation and region, without emphasizing the legal form.
- Show income for one year or several?You show income for the same period you benchmark against - usually the last year or the last few years in matching slices. The community notes you should compare against statistically high salaries, not just the average, and that summing several average incomes from different jobs is a weak strategy.
- Do bonuses and RSU count toward the salary criterion?Community opinions differ: formally you compare remuneration, but the comparison must be apples to apples. If market statistics show base salaries, compare your base pay; if you have a tax form with gross income, officers look at the whole figure. The key is an identical basis on both sides of the comparison.
- Does a team or company award count?By default USCIS expects an individual award, and team awards often get discounted. But the community has seen successful cases where the applicant proved the award happened because of them - letters from the organization, a description of their personal role. The alternative: file a team award under contribution or leading role instead of Awards.
The occupation in the U.S. Department of Labor classification (O*NET)
Data Scientists O*NET-SOC 15-2051.00closest related occupation
- Median wage (BLS, 2025)
- $120,230a year · $57.8/h
- Projected growth (2024-2034)
- much faster than average(7% or higher)
- Projected openings (2024-2034)
- 23,400
The occupation page on O*NET OnLineThe high-salary criterion: how it is comparedThe approved EB-1A template files this page as Exhibit 1.1
Source: O*NET OnLine (U.S. DOL), wages BLS OES 2025; the median across all U.S. workers in the occupation. Checked 2026-09-09.
From community practice
For a leading or critical role, members advise proving it with letters that describe your impact on the organization's outcome, not just your job title.
High remuneration is judged against the market for your profession and region; discussions advise attaching industry salary surveys as a benchmark.
A talk at a selective conference supports original contribution, and the published proceedings support authorship; an invited or keynote talk is valued above a poster in discussions.
Distilled from public community chats - observations, not legal advice.
From community discussions
«I fully support this. Essentially, NIW is about the profession. They want ML, no problem. But EB1? It should deeply matter less what you do, the key is how impressive it is. In contrast, I've seen such nonsense in ML that it's frustrating that many take advantage of this hype.»
community member · from public community chats
«I was looking for a job for O-1 about six months ago (so they could act as a petitioner), while being outside of Russia, without permission. I received an offer in about four months. Of course, the salary is not at the U.S. market rate but at the market rate of the country I'm in, but I managed to sign the paperwork. I personally know another person who searched for about the same amount of time and also received an offer and was ready to sign all the documents. In general, I'm saying that while it's difficult, the field of ML is achievable.»
community member · from public community chats
«I'm sending an update - after 19.2 months of administrative processing, I finally received my O-1 visa in São Paulo (I updated the table). The actual processing took about 14-16 months, then I just couldn't get to Brazil to submit my passport. The consulate didn't say until the last moment why they were requesting passports and what the status of the review was, so I flew by chance and luckily it worked out. My specialty was - Machine Learning Engineer.»
community member · from public community chats
«I have a Telegram chat with 16k people about writing resumes for IT folks. I myself plan to apply as a software engineer. The AI chat suggests stuffing this into judging the work of others or into original contributions / community leadership. Any ideas on how to tie this activity to the case?»
Artem · from public community chats
Personal opinions of community members from public discussions, not legal advice.
What the chats ask about the ML engineer profession
- 1×I have a master's degree in Computer Vision and work as a programmer. Is EB2-NIW suitable for me?
- 1×How do USCIS officers understand that the petition text is written using neural networks? How to avoid this?
Where people go wrong - from the chats
- mistakeNot every achievement qualifies for a talent visa; it's important to have significant accomplishments and consider managerial experience.
- mistakeThe lack of clear classification of specialties can complicate the visa application process, especially in specific fields.
- mistakeCompetition in interviews for ML engineers can be very high, especially in areas like AI/ML in fintech.
- mistakeGeneral skills may not be sufficient for a successful visa application; it's important to emphasize specific achievements.
Cases from the community
EB-2 NIW · IT and machine learning · outcome not stated
A candidate with a combination of economics, legal, and technical experience, as well as extensive tenure in the transportation industry, was considering a machine learning project.
What they did: It was suggested that the argument for waiving labor certification should be built around interdisciplinary expertise, industry experience, and the project's potential benefit to the economy.
Questions and answers on the site
Distilled from the public community chats, anonymised; corpus as of June 2026. These are participants' observations, not legal advice and not USCIS statistics.
Chats in your field
Community chat
Rating 51IT and software on EB-1: open-source, talks, articles and leading role for technical specialists.
Knowledge base
Frequently asked questions
- Which talent visa fits for an ML engineer?
- Yes, an ML engineer is one of the strongest IT profiles for the talent visa. The usual criteria are original contribution (models, open source, AI infrastructure used by others), high remuneration (typical for ML/AI roles when benchmarked against the market), judging (paper review, hackathons, grants) and a leading role. A title alone is not a criterion - you need documented impact beyond routine engineering work.
- How many criteria do you need to meet?
- At least three (or one major international award); for an ML engineer the usual ones are "Original contribution", "High remuneration", "Judging", "Leading role", and "Original contribution" was accepted in 16% of the AAO appeals that claimed it; EB-2 NIW uses the three Dhanasar prongs instead.
- Can app downloads or sales through platforms be used as evidence?
- Yes, downloads and purchases through electronic platforms can qualify if you can prove them: for example, the App Store or book services provide sales and download reports through the account dashboard. The main thing is to confirm the figures with official exports. Where exactly to place this argument (commercial success or contribution) depends on the profession and the case.
- Does software copyright count as a contribution if a patent takes too long?
- Copyright is simpler, faster and cheaper to register than a patent, and the community has filed it as part of the evidence. But, like a patent, the mere fact of registration is weak - you need to show use and significance (recommendation letters from companies, adoption). The strongest is the combination: the intellectual property object plus proof of its application.
- Can a recommendation letter from a scientific editor replace a review?
- Yes, in the community's experience a recommendation letter about your article can be even more useful than a review. For example, people choose an authoritative editorial board member of an international journal where you have published and ask for a letter of recommendation - the result reads like an analysis of your articles plus an assessment of your contribution. This strengthens both authorship and contribution.
- Which filing mistakes do the chats name for an ML engineer?
- Not every achievement qualifies for a talent visa; it's important to have significant accomplishments and consider managerial experience. The lack of clear classification of specialties can complicate the visa application process, especially in specific fields. Competition in interviews for ML engineers can be very high, especially in areas like AI/ML in fintech.
- What do the chats say about the talent visa for an ML engineer?
- «I fully support this. Essentially, NIW is about the profession. They want ML, no problem. But EB1? It should deeply matter less what you do, the key is how impressive it is. In contrast, I've seen such nonsense in ML that it's frustrating that many take advantage of this hype.» - a community member, from the public chats
Other professions in "IT / software"
Understand the terms and compare visas
From the chats and the documents
Documents and templates from the library
BCS Fellowship interview prep guide, Apr 2026Guide7 pages anonymized access on request
NIW Exhibit: Principal Investigator on Two ProjectsTemplate1 page anonymized access on request
USCIS Proposed Rulemaking - Employment-Based Immigrant Petition ReformsGuide1 page anonymized
Checklist - O-1 / EB-1A / EB-2 NIWGuide1 page anonymized access on request
Reference information and community experience, not legal advice and not a USCIS decision. The AAO figures are lessons from appeals, not approval odds; verify on uscis.gov.