EB-2 NIW for Software Engineers: How to Build a Winning Self-Petition (No PhD Required)

Maybe you read our deep-dive on proving EB-1A original contributions and thought: solid framework, but my open-source project has hundreds of users, not hundreds of thousands. My work matters, but "sustained national acclaim" isn't the phrase for it yet.

That thought is exactly who the EB-2 National Interest Waiver exists for. The NIW asks a fundamentally easier question than EB-1A. Not "are you one of the top people in your field?" but "is your work important to the US, and are you well positioned to keep doing it?" No acclaim requirement. No employer, no PERM, no job offer. And despite the folklore, no PhD.

One thing before the optimism: the NIW of 2022, where a STEM degree and a paragraph about AI practically self-approved, is gone. Approval rates fell from roughly 96% in FY 2022 to 55% in FY 2025, and for one quarter denials actually outnumbered approvals. The petitions still winning share one trait, and this article is about building it: specificity.

The No-PhD Math

EB-2 has a baseline qualification, and engineers routinely clear it without noticing. You need either an advanced degree or its equivalent - and the equivalent is a bachelor's degree plus five years of progressive post-degree experience.

"Progressive" means increasing responsibility: engineer to senior to lead, growing scope, harder problems. A senior engineer seven years out of undergrad typically qualifies on experience alone. A master's in CS or related field qualifies outright. The PhD assumption comes from NIW's researcher-heavy history, not from the regulation.

There's also a second door, exceptional ability (three of six criteria like ten years of experience, high salary, and recognition), but most engineers take the degree-plus-experience route because it's cleaner. The step-by-step filing guide covers the mechanics.

Dhanasar in One Screen

Every NIW is judged on the three-prong test from Matter of Dhanasar (2016). In plain English:

1. Your proposed endeavor has substantial merit and national importance. The work matters, and it matters beyond your employer and your city.
2. You are well positioned to advance it. Skills, track record, and momentum say you can actually do this.
3. On balance, the US benefits from waiving the job offer and labor certification. Requiring a PERM here would cost the country more than it protects.

All three, not two of three. We've broken down the full legal framework separately; this article is about making each prong true for engineering work.

The Endeavor Is the Whole Game Now

In January 2025, USCIS updated its policy manual with guidance that applies to every NIW pending or filed since. The operative change: your proposed endeavor must be a specific undertaking with defined goals and methods. An occupation is not an endeavor. A field is not an endeavor.

That update is the quiet explanation for the approval-rate collapse. "I am a machine learning engineer and AI is important to America" was approvable in 2022 and is a template denial in 2026.

Compare these:

Fails prong one: "I develop software in the artificial intelligence field, which the US government has identified as critical."

Survives prong one: "My endeavor is building open-source observability tooling for large-scale ML training clusters, reducing failed training runs and wasted compute for US organizations - measured by adoption across N production deployments - with a roadmap targeting the reliability failures documented in [specific industry problem]."

The second one has a what, a for-whom, a measurable dimension, and a forward plan. Notice it's an endeavor you can pursue across employers - that's the point of a self-petition. Your day job can be evidence of the endeavor without being its boundary.

Four National-Importance Framings That Fit Real Tech Work

Prong one has two halves: substantial merit (almost never the fight) and national importance (always the fight). For engineers, the strongest cases anchor the endeavor to a documented federal priority, then prove the endeavor itself - not just the field - carries national weight.

AI infrastructure and reliability. The federal government spent 2025 putting AI at the center of national strategy - an AI Action Plan in July, executive orders accelerating data center and compute buildout. If your endeavor is the unglamorous substrate (training infrastructure, inference efficiency, evaluation and safety tooling, data pipelines), the national hook is real and current. Your job: connect your specific layer to outcomes - compute costs, reliability, capability the US side gains.

Cybersecurity and critical infrastructure. The US formally designates sixteen critical infrastructure sectors, and software security for energy, finance, healthcare, and water systems sits squarely in national-security territory. Endeavors in vulnerability research, secure-by-design tooling, supply-chain security (the software kind: dependencies, build systems, provenance), and incident response tooling map cleanly. Evidence: CVEs credited, adoption of your security tooling, standards participation.

Healthcare data and systems. Interoperability, clinical data security, and outcome-improving analytics touch both a massive share of the economy and explicit federal modernization pushes. The framing that works is patient-level and system-level impact: what your data platform or algorithm changes about care delivery, cost, or safety - with numbers from deployments.

Semiconductors and the chip supply chain. Domestic chip capacity has been a bipartisan priority since the CHIPS Act, reinforced by 2025 orders fast-tracking fabs and their surrounding infrastructure. This extends beyond silicon engineers: EDA tooling, yield-analysis software, fab automation, firmware for domestic hardware, and the simulation stack all serve the buildout.

Prong Two: Well Positioned, Engineer Edition

Prong two is where your record lives, and here's the efficient part: the evidence structure is a scaled-down version of what the EB-1A original contributions criterion demands. Same artifacts, lower bar. You don't need the field to have changed because of you; you need credible momentum.

What officers weigh:

• Education and experience matched to the endeavor specifically, not to software generally. The January 2025 guidance made this explicit for exceptional-ability petitions, and the same logic runs through prong two for everyone: officers want your qualifications tied to the thing you say you're going to do.
• A track record of the actual work. Shipped systems, deployments, open-source projects with real users, patents, publications or serious technical writing, conference talks. Metrics beat adjectives: adoption counts, scale carried, problems retired.
• Letters that attest with data. The same rules from the EB-1A world apply - independent voices, checkable facts, no template phrasing. Current expectations run quantitative: engineers cite adoption and deployment numbers the writer can vouch for.
• A forward plan with progress. The endeavor needs a roadmap and evidence you're already moving: commits, pilots, users, collaborations. "Well positioned" is a trajectory argument, not a resume summary.

Prong Three: Why Skipping PERM Helps America

Prong three asks why the US should waive the labor certification that normally protects the domestic workforce. For self-directed technical endeavors, the honest arguments are structural:

PERM certifies one job at one employer - and your endeavor spans employers by design. A labor market test can't evaluate open-source infrastructure or cross-industry security work. Timing matters when the endeavor serves an active national push; PERM's multi-stage timeline delays the benefit. And your qualifications exceed what any single job posting would capture.

Officers respond to prong three arguments that are specific to your situation rather than recited. One strong paragraph connecting your endeavor's urgency to the impracticality of a labor certification beats a page of boilerplate.

The Honest Part: Approval Is Not a Green Card Date

An approved NIW gives you an I-140 you own - no employer can withdraw it, and your priority date is locked. What it doesn't do is move the visa bulletin.

NIW sits inside EB-2, so an approved NIW joins the very same EB-2 queue as every PERM-based filing from your country - and EB-2 India is currently unavailable outright for the rest of the fiscal year. If you're India-born, an NIW approved today is a place in a very long line, not a near-term card.

So why file anyway? Because the approved I-140 is leverage that compounds: it locks your date while you're free to change jobs, it supports H-1B extensions past year six, and it runs in parallel with anything else - a later EB-1A with a ported priority date, an employer's PERM, whatever your situation adds. The NIW-vs-PERM math still favors owning your petition.

Surviving the 2026 RFE Climate

The numbers say prepare for contact: case-tracking data put NIW RFE rates around half of filings in January 2026, easing to just under 40% by March, with approval rates recovering into the mid-40s percent after the late-2025 trough. An RFE is now a normal event, not a verdict.

Three habits separate the petitions that clear it:

One endeavor, told once, everywhere. Your statement, letters, and exhibits should describe the same specific endeavor in the same terms. Officers flag files where the resume says backend engineer, the statement says AI researcher, and the letters say visionary.

Evidence that survives checking. Adoption numbers with sources, letters whose claims a phone call could confirm, no inflated metrics. Assume verification; round down.

Address the weak prong preemptively. Most engineer NIWs are strongest on prong two and thinnest on prong one's national-importance half. Spend your best pages there - the bridge from your work to field-level effect - because that's where the RFE will aim.

FAQ

Bottom Line

The NIW is the self-petition with the accessible bar: bachelor's plus five years of progressive experience clears the degree hurdle, and the test is importance and positioning, not acclaim. What 2025 changed is the tolerance for vagueness - a specific endeavor with defined goals, anchored to a real federal priority and bridged to field-level effect by your own evidence, is now the entire difference between the 55% who get approved and the rest.

Build the endeavor statement first, make every letter and exhibit tell that one story with checkable numbers, and treat the national-importance bridge as your most important pages. And if your record keeps growing, remember the two-lane play: the EB-1A evidence you're accumulating and this petition share a spine, and the priority date you lock today travels with you.

Want to see how NIW stacks against EB-1A, PERM, and every other route for your specific country and timeline? The comparison tool runs them side by side with current bulletin data, and the petition prep template gives you the evidence inventory structure for either self-petition.