NIW for Data Scientists: How USCIS Actually Categorizes Your Case

"Data scientist" is not a USCIS category

Data scientist is one of the most common job titles in tech, but USCIS does not have a line item for it. The agency adjudicates the proposed endeavor and the evidence, and when the AAO publishes appeal decisions it sorts them into broad occupational groups. Data science work falls under Information Technology & Computing — the same group as software engineers and IT specialists.

That matters because there is no honest "data scientist NIW approval rate" to quote. Any guide that gives you one invented it. What we can show you is the IT & Computing appeal record, which is real, and how data-science petitions tend to succeed or fail inside it.

The IT & Computing appeal record

Our database holds 6,362 AAO decisions on EB-2 National Interest Waiver appeals from 2015 to 2026. Each one is an appeal — a petition USCIS already denied. Across the full set, 375 appeals were approved (5.9%), 5,292 were dismissed (83.2%), and 533 were remanded (8.4%).

Within Information Technology & Computing, the AAO sustained 77 of 1,069 appeals — a 7.2% rate. The narrower slices we can isolate:

• Information Technology: 11 of 185 appeals approved (5.95%)
• Software Engineering: 5 of 74 appeals approved (6.76%)

These are appeal-stage numbers. First-pass USCIS approval is much higher and is not in this dataset. The appeal record is a map of how thin petitions get rejected, which is exactly what you want to study before you file.

One useful contrast: Scientific Research petitions — physical and life sciences — are sustained on appeal at 19.3% (93 of 481). A data scientist whose work is genuinely research, published and cited, can be read against that pattern even though the petition files under computing. The classification follows the evidence, not the job title on your offer letter.

The Prong 1 problem for data scientists

Across the appeal record, dismissed petitions cluster on Prong 1 — substantial merit and national importance. For data scientists this is the hardest prong, and the reason is structural. A lot of data-science work is, in plain terms, commercial: optimizing ads, building recommendation systems, business intelligence for a single company. That work can be valuable and still fail Prong 1, because national importance is about benefit to the country, not to an employer.

The petitions that clear Prong 1 connect the work to a documented public need rather than a private one.

Profiles that tend to clear Prong 1

• Health and clinical data science. Disease prediction, public-health surveillance, clinical decision support. Work with EHR or epidemiological data at research institutions or public-health agencies has a clear national-benefit story.
• Environmental and climate data. Climate modeling, environmental monitoring, energy-grid optimization — all map onto stated federal priorities.
• Government and policy data science. Work for federal agencies, government contractors, or academic groups doing policy-relevant analysis.
• Published methodological research. A data scientist who publishes peer-reviewed work that other researchers cite is in a far stronger position than an industry-only practitioner.

Profiles where Prong 1 is an uphill fight

• Recommendation systems and advertising optimization
• Business intelligence and analytics for a single commercial company
• Generic "data-driven decision-making" framed only around private revenue
• Data engineering and pipeline infrastructure, even at large scale

If your current job is on the second list, you are not disqualified. You need to find the public-benefit thread in your work — or in your published research — and make that the spine of the petition rather than a footnote.

Building a Prong 1 argument that survives

Connect your specific work to a documented national need. Adjudicators have read "data is important" thousands of times; it persuades no one. Concrete anchors that do:

• Your models are used in clinical decision support that is FDA-regulated or clinically validated
• Your work addresses a specific disease burden or a named public-health gap
• Your environmental models feed EPA, NOAA, or state regulatory decisions
• Your research was funded through competitive NSF or NIH grant review
• Your published methods are used by researchers at multiple institutions

Prong 2 evidence specific to data science

Prong 2 asks whether your record predicts you will advance the endeavor. Data scientists have a few evidence types worth knowing about:

• Published datasets. If you created and released a dataset that others use, citations of the dataset paper are strong, direct evidence of field impact.
• Citations read in context. A citation count is a starting point. Show who builds on your work and how. Our citation map tool traces where your citations come from, which helps argue influence beyond your own institution.
• Conference contributions. KDD, NeurIPS, ICML, or domain venues like AMIA for health informatics. State the venue and its selectivity.
• Kaggle and competition placements. Top finishes signal peer recognition, but adjudicators respond unevenly to them. Treat competition wins as supporting evidence, not as the centerpiece.

The career-transition issue

Many data scientists file mid-career, after moving from academia into industry. The academic work from three to five years ago may be strong, while the current industry role is harder to frame for national interest. The approach that holds up: build the petition around your continuing research contributions, even if they are now part-time or side projects, and around the specialized expertise you carry that a labor-market test would not readily surface.

What to do before filing

1. If your work has any public-benefit dimension, make it the spine of the petition, not a footnote
2. Pursue competitive federal grant funding before filing if you can — it strengthens both Prong 1 and Prong 2
3. Publish your methodology and release your models or datasets where possible
4. Get independent letters from academic researchers who actually use your methods or data
5. Run a case review to compare your profile against real AAO decisions before you commit

For the full IT & Computing breakdown, see the profession lookup tool. When you are ready to draft, the petition builder walks through each Dhanasar prong in order.