He Rewrote the Algorithm. Then He Rewrote His Life.

I first met Sunny Jaiswal on a Tuesday afternoon in 2022. He had been referred to me by a colleague, and when he walked into my virtual office carrying a folder of papers he had color-coded with sticky tabs, I could see immediately that this was a man who had done everything right and still felt like he had done everything wrong.

He had been in the United States for nine years on an H-1B visa. He had a machine learning patent filed in his name. He had been cited in over forty academic papers. He had built the recommendation engine powering the content feed of a mid-size social platform used by more than thirty million people. By any reasonable measure, Sunny was extraordinary. But the immigration system had twice told him otherwise.

The first rejection - a lesson in framing

Sunny's first EB-1A petition was filed in 2019 by an immigration attorney who specialized primarily in corporate immigration. The petition was thorough. It listed his publications, his patent, his speaking engagements at two regional tech conferences. What it failed to do was contextualize any of it. USCIS is not a room full of AI researchers. An adjudicator reading about "contrastive learning loss functions for sparse recommendation graphs" has no way to understand why that work matters unless someone takes the time to explain it in plain, demonstrable terms. That petition listed accomplishments. It never argued significance.

The Request for Evidence arrived six months later asking for more documentation. The supplemental response was filed. The denial came three months after that. The reasoning was clinical: insufficient evidence of sustained national or international acclaim. Sunny read it four times that evening. He said it felt like being told the sun was not bright enough.

The system does not reward brilliance. It rewards brilliance that has been correctly translated into the language the system understands.

The second attempt — closer, but still not enough

In 2021, Sunny tried again with a different attorney and a significantly stronger dossier. In the two years since his first filing, his professional profile had deepened considerably. He had joined a Series B startup as their Head of AI Infrastructure, where he led a team that reduced model inference latency by sixty-three percent across their production stack. His citation count had climbed past one hundred. He had been invited to peer-review submissions for a top-tier NeurIPS workshop.

The second petition was better argued. It leaned into the judging criterion and his high salary relative to others in the field. It gathered letters of support from two well-known professors at Carnegie Mellon and one from a principal scientist at a major tech company.

USCIS denied it again.

This time the reasoning was cut differently. They acknowledged the peer review work. They acknowledged the salary evidence. But they questioned whether his contributions had risen to the level of "major significance" in the field. The letters, though well-credentialed, had been too general. They praised Sunny's intelligence and character. They did not explain, with any specificity, what had actually changed in the field because of him.

When Sunny called me after that second denial, there was a long silence before he spoke. He said he was thinking about abandoning the EB-1A category entirely and filing for a national interest waiver instead. I understood the impulse. After two rejections, the EB-1A can begin to feel like a door that simply does not open for you. But I had read both of his prior petitions carefully. The issue was never his eligibility. The issue was always the argument.

Rebuilding from the foundation

When we began working together, the first thing I told Sunny was that we were not going to file for at least six months. He looked almost offended. He had been waiting for years. Six more months felt like punishment. I told him I understood and that those six months were going to be the reason we won.

We started not with documents but with a conversation. I asked him to walk me through every project he had worked on in the past five years. Not the way he would describe it on a resume. The way he would describe it to a friend over dinner. No jargon, no titles, just the story of what he had tried to do, why it was hard, and what happened when it worked.

What emerged was something neither of his prior petitions had captured: a through-line. Sunny had spent a decade working on a specific, identifiable problem. The cold-start problem in sparse recommendation systems, where models perform poorly for new users or new items because of insufficient interaction data. This problem affects every recommendation platform in the world. His original approach to it, using a form of contrastive pre-training adapted from natural language processing, had been novel enough to attract citations from researchers at Google, Meta, and Spotify. He had not just published a paper. He had introduced a technique that others had built upon.

That was the story. That was the argument.

The architecture of a winning petition

Over the following months, we built what I call a triangulated argument. One that establishes extraordinary ability not through the volume of evidence but through the convergence of multiple independent lines of proof, all pointing to the same conclusion.

We restructured his citation evidence entirely. Rather than simply listing the number of citations, we identified the twenty-three most significant ones, those from researchers at institutions with the strongest reputations in machine learning, and wrote a detailed analysis of each one. We explained what the citing paper was trying to do, where it referenced Sunny's work, and what that reference actually signified. We showed that researchers were not merely acknowledging his paper's existence. They were building on it. One team at a European university had specifically used his pre-training framework as a baseline for their own work, which was subsequently cited over three hundred times. We made that connection explicit and fully traceable.

For the support letters, we replaced the previous approach of general letters from impressive people with highly specific, technically substantive letters from people in the best position to describe the real impact of Sunny's contributions. We worked closely with each letter writer on the substance to ensure they addressed the right questions. What did the field look like before this contribution? What changed because of it? Why does that matter to the broader trajectory of the discipline?

One of the most powerful letters came from an engineer at a mid-size streaming company who had implemented Sunny's technique at their platform and documented, in measurable terms, the improvement in new user retention that followed. That letter was not from a professor. It was not from a luminary. But it was devastating in its specificity, and specificity is what wins.

We also addressed the prior denials directly and strategically. Many attorneys advise against referencing prior denials in a new filing. I disagree. When a petitioner has been denied twice, an adjudicator who pulls the file will see those denials. The question is whether you want to explain them on your own terms or leave the impression they left standing on their own. We wrote a brief, factual explanation of what each prior petition lacked and why this filing had addressed those gaps. We did not argue with USCIS. We acknowledged the deficiencies and demonstrated that they had been remedied.

The wait that almost broke him

The petition was filed in March 2023 under premium processing. Within fifteen days, we received not an approval but a Request for Evidence. My heart dropped for a moment. Then I read it carefully. The RFE was narrow. USCIS was asking for additional evidence on two specific criteria: the judging criterion and the critical role criterion. They were not questioning the foundational argument. They were asking us to deepen two specific threads of proof.

I called Sunny. He had been through two denials. An RFE felt to him like the first sign of a third. I told him the opposite was true. A narrow RFE on two criteria, after a premium processing filing, meant that the adjudicator had found the rest of the petition persuasive. They were not rejecting us. They were asking us to close two remaining gaps.

We filed the RFE response within thirty days. For the judging criterion, we obtained documentation from three additional workshops and conferences where Sunny had served as a reviewer, including one where he had been specifically invited to join the program committee based on his expertise in the cold-start problem. For the critical role criterion, we obtained an expanded declaration from Sunny's CEO, supplemented by an internal performance review, a board meeting presentation that referenced his infrastructure work by name, and a letter from the company's lead investor specifically crediting his technical leadership with enabling a successful Series C fundraise.

We also requested a third-party technical expert declaration from an AI researcher at a respected university who had no prior relationship with Sunny. He reviewed the body of work and provided an independent assessment of its significance. That declaration took three weeks to obtain. I believe it was the piece that closed the case.

An EB-1A is not a reward for excellence. It is an argument. The excellence must be real, but the argument must be built with intention.

The approval

On a Thursday morning in November 2023, I received the email before Sunny did. His I-140 had been approved.

I sent him a message that said only: "It's done."

He called me eleven minutes later. He did not say much at first. He asked me to confirm what I had said. I read him the approval notice. There was silence, and then a sound I will not try to describe here. It was the sound of a person who has been holding their breath for four years finally exhaling.

He told me that his mother, back in Pune, had been lighting a lamp at a small temple near their house every Thursday for the past eighteen months, ever since the RFE arrived. He said he was going to call her next.

I told him to go. The debrief could wait.

What Sunny's story taught me and what it can teach you

I have worked on hundreds of EB-1A cases across technology, science, business, and the arts. Sunny's story is exceptional in its specifics but entirely representative in its shape. The majority of EB-1A denials I have seen were not denials of merit. They were denials of argument. Petitions that told USCIS what a person had accomplished without ever telling them why it mattered, who it influenced, and how those influences could be demonstrated with evidence rather than just claimed with words.

The EB-1A standard is genuinely high. USCIS is not wrong to demand proof of sustained national or international acclaim. But acclaim does not document itself. A researcher with a hundred and twenty-three citations and a granted patent is not automatically EB-1A-worthy on paper. The petition must show the adjudicator, step by step and criterion by criterion, what those numbers mean in the context of the field, what level of recognition they represent relative to peers, and why the contributions behind them constitute a significant original contribution to the discipline.

That translation from lived achievement to legal standard is what immigration strategy is actually about. It is the work I do every day at Visa Architect. Done carefully and honestly, it can turn the record of an extraordinary life into the argument that extraordinary life deserves.

If you are a tech professional navigating the EB-1A or wondering whether you even qualify, I would love to speak with you. A strong professional record and a strong petition are not the same thing. But with the right strategy, they absolutely can be.

Every case I take begins with the same question Sunny and I started with. Not "do you qualify?" but "what is your story, and how do we tell it so the right people truly understand it?"

The answer to that question is what wins.

Swatilina Barik

Founder and Principal Strategist, Visa Architect