80 Days to Stay
When mass layoffs put H1-B holders on a 60-day clock, most funded startups that could sponsor them don't even know the option exists. This project uses public SEC filings to close that gap.
Mass layoffs at major tech and finance companies create a specific, brutal deadline for one group of employees: internationals on an H1-B visa who lose their job get 60 days to find another one that pays roughly what they were making, in a job market where AI is already disrupting hiring across the board. 80 Days to Stay is a new Humanitarians AI initiative built in direct response to that problem, and it starts with a simple bet: the jobs these people need already exist, they're just hidden.
The bridge Humanitarians AI already builds
Humanitarians AI's existing purpose is to serve as a bridge, letting people continue doing serious AI research and continue their education through volunteer work, up to a year, until they land a strong job. Without that bridge, someone would face a 90-day window to find work in a market where AI is disrupting hiring across the board. This new problem, the 60-day clock facing laid-off H1-B holders at large companies, is a variation on the same underlying issue, arriving through a different door.
Where the 80 days comes from
Companies doing layoffs typically give a couple of weeks' notice, which is standard practice, and then the affected employee has 60 days after their last day to find qualifying work. Add those together and you get roughly 80 days total, which gives the project its name, with an intentional nod to Around the World in Eighty Days as well. The framing here is a kind of public, daily working log: what can actually get built, and shipped, in that same 80-day window, aimed at helping this specific group of people.
The real problem: two groups that aren't connecting
The read on the underlying problem is specific. Large, well-known companies, the Microsofts and Goldman Sachses of the world, have entire teams that understand visa sponsorship and H1-B processes. But those companies make up a small slice of total US employment, maybe one to three percent of the economy, not the majority of available jobs. Meanwhile, AI has driven a wave of startup funding, with companies raising anywhere from one million to over a hundred million dollars and needing to hire aggressively. The problem is that a large share of these funded startups avoid hiring internationally not because they've decided against it, but because visa sponsorship is unfamiliar territory to them, and the actual requirement is mostly paperwork and straightforward compliance that low-cost services can handle on their behalf. Talented people need jobs. Funded startups need talent. The two sides simply aren't connecting, and there isn't an existing tool built specifically to surface this hidden market, as opposed to the many tools that just aggregate listings from companies like Microsoft or Tesla that are already well known.
The plan: a searchable database of funded startups
The proposed fix is a free, searchable platform connecting funded startups with talent, including international talent, and giving those startups direct information about the visa process along with pointers to services that can handle the paperwork for them. It's explicitly designed to work for everyone, not just international candidates, while giving those candidates, whose presence already helps fund US universities, another real avenue beyond the small set of companies that already understand visa sponsorship.
Day one: finding the data
Every version of this idea starts with the same question: where do you get reliable data on which startups actually have funding? The answer turned out to be public record. Any US company that raises money from investors, including angel investors, is required to file with the SEC, specifically a Form D filing, even as a private company. That filing discloses the company's name, address, and how much money it raised. Full public-offering disclosures aren't the target here; the funding data alone is enough to identify which companies plausibly have the resources to hire.
Because it's a government filing, companies have a strong incentive not to misstate it, and larger companies, the ones with a million dollars or more in funding, are especially unlikely to skip the filing given the compliance risk. The companies most likely to miss filing tend to be very small ones that simply aren't aware of the requirement, since they don't have a legal team flagging it for them, which is a reasonable trade-off for a project focused on startups with real hiring capacity.
Why this beats existing data services
Commercial alternatives exist, but they're framed here as expensive relative to what they actually deliver, on the order of $99 a month for the ability to look up a handful of companies and export a small spreadsheet, or a paid LinkedIn Sales Navigator-style subscription. SEC Form D data, by contrast, is free, public, and comes directly from an investor-verified government filing rather than a scraped or inferred source. Getting it is mechanical: the SEC's website provides downloadable XML files that just need to be parsed into a more usable format like TSV or JSON.
Where this goes from day two onward
Day one's task was straightforward: identify essentially every funded startup in America with a name, an address, and its principal people, using nothing but data the SEC already makes public. The plan from there is layered enhancement: converting the raw XML into structured formats, and eventually using AI to make inferences, for example about whether a company already hires internationally, based on signals like the composition of names on their team. That enrichment work is explicitly framed as a strong hands-on project for anyone interested in data science, since it's fundamentally about taking a raw dataset and progressively enhancing it into something genuinely useful.
The build: free tools, zero budget
The technical stack is deliberately unglamorous and cost-free: Python, React, FastAPI, and PostgreSQL, all on free tiers, with a stated monthly budget of zero dollars. The project is being run directly as a nonprofit initiative, open to volunteers who want an OPT-eligible project with direct supervision, meeting with the project lead a couple of times a week.
Key takeaways
- H1-B holders laid off from major companies face a 60-day window to find qualifying work, and with standard notice periods added in, that adds up to roughly 80 days total.
- Well-known big tech and finance companies represent only a small slice of total US employment, while a large wave of AI startup funding has created hiring demand these companies don't capture.
- Many funded startups avoid hiring international talent not by choice but because visa sponsorship is unfamiliar to them, even though the actual process is manageable paperwork and compliance.
- SEC Form D filings are free, public, government-verified data on which US companies have raised real investment funding, making them a strong foundation for identifying startups that can plausibly sponsor hires.
- The build uses an entirely free tech stack, Python, React, FastAPI, and PostgreSQL, with a $0 monthly budget, and is run as a direct, supervised nonprofit project open to OPT-eligible volunteers.
Who this is for
This is for data engineers, full-stack developers, and researchers who want to work on a concrete, visible project with real stakes, no prior experience required beyond 5 to 10 hours a week and genuine commitment. It's also directly relevant to international students and H1-B holders trying to understand where the hidden opportunities in the current job market actually are. The project is run through Humanitarians AI with direct supervision for anyone treating it as an OPT project.
Full transcript(auto-generated, with timestamps)
[0:00]Okay, bear here. A new initiative, humanitarians AI. So for you people who do not know, uh, humanitarians AI serves as a bridge. We allow students to continue to do research, to continue to learn, to continue their education until they get that great job. I'll have other videos on the importance of international students to funding universities. They are the lifeblood of universities. I'll go through other videos talking about that, talking about how they give opportunities to American students because of the funding they bring in, but that'll be a separate video. So, what's happened recently is so the purpose of humanitarians is to uh allow people to continue to do
[0:40]Research uh because otherwise you'd have a 90-day window to find a job. And in this job market with AI just disrupting everything, it's just difficult. And so, we serve as a bridge. people can do a volunteer up to a year until they get that great job. And the p purpose of humanitarians AI is to allow people to continue to learn to do serious AI research until they get that great job. But now there's another issue which I just real I didn't realize but some friends of mine told me about. So with the AI changing a lot of things, it's led to a lot of mass layoffs in a lot of
[1:17]Big tech companies, big tech, big finance, you know, the the Deoids, the the Microsofts, the the Goldman Sachs, the that group. So what happens is that puts internationals in a in a particularly difficult spot because what happens is if they're no longer with the company and they're under H1 visa, they have 60 days to find another job which pays sort of roughly what they've been making, which now in this case is a fairly high bar because they're working at sophisticated companies for sophisticated jobs. So this has happened to a few people that I know and so my sort of response to things is to try to
[2:00]Fix them. So here's my approach to do something about this issue. It's called 80 days to stay. The 80 days comes from a lot of these companies give people a couple weeks notice which is proper and their last day is you know in a couple weeks and then they have 60 days after that couple weeks. So this turned out to be 80 days. 80 days is also reminiscent of one of my favorite books 80 days around the world. This is a brilliant but and so this is my Phyious vlog attention. So what can I do in 80 days to help this group of students? So
[2:37]Here's day one. So the problem is the problem the way I see it and any international student or anybody who understands this world is more than uh welcome to comment on this video the problem is I see it is the big companies the the Teslas the Microsofts the Goldman Sachs the deoids they they all know this world they have entire teams that that you know know H1 stuff uh do stuff but there are many that they're not you of all of the economy. They're maybe 1 or 2% of the total economy or maybe 3% of the US. Actually, Google may be a little bit more because now these
[3:18]Are now trillion dollar companies, but they're not everything in terms of jobs. They're not the most jobs. And again, I'll do another video on exactly what portion of the economy are these big companies. And but AI, what AI is doing is exploding a lot of entrepreneurship, a lot of funding for AI startups. But a lot of these startups have money. They've raised a million, five million, 10 million, 50 million, $100 million, and they need talent. They need to find the best talent, but a lot of them shy away. So even though they have the money is very common and this is the research I done
[4:00]Something like 90% of those companies I think it's in here somewhere but whatever the percentage is the I think the percent is on the GitHub where where where um they they just avoid this world because they don't understand it it's foreign to them so that the basic is they they don't really understand the reality of it's basically filling out a bunch of forms and then complying to some basic rules and they're not that hard. There are many services that will just do that for them and it's not if they have any in-house stuff they can do it but if not they can hire services which will handle
[4:36]That at a low cost but because they don't have the knowledge and they don't really understand this world. Uh a lot of and I'll again I'll do another video on the exact numbers because I've done research on those numbers. A lot of these startups that need desperately need tech talent are not tapping the source of talent. And so there's a problem both ways. The the people with the talent need the jobs. The startups need the talent, but they're not connecting. And so I don't think we need to create another tool which helps people just look up jobs at Microsoft or Tesla or whatever. There there are bajillion of them out
[5:17]There. But I haven't seen anything out there that is looking for this hidden market that is connecting talented people with AI startups. And so this is the idea. So the idea is to create a database of basically every startup and I'll day one I'll talk about in a second how to do that. But as everything everything starts with the data. We need to get the data. Who is who are they coming? where the company is, how much funding they have. So if they don't really have any funding, if they have $50,000, they're not going to hire anything. Now they have a million, 10 million, 15 million, 100 million.
[5:59]They need people. So our idea is to create a little searchable website which is looking for connecting these startups with talent and particularly international talent including giving them information about the the visa process and the companies that could basically just handle that for them. It's not that difficult. But again, most companies just because they don't know this world will just shy away from it. So what I'm going to do is every day do something toward this goal. Making a website which allows um startups, venture firms to connect with the best talent, whether they're American or Australian or Indian or whatever. So it works for everybody. And
[6:58]What this does for the international students which basically fund US universities. It gets them another opportunity there. There's nothing that prevents them from still applying to Tesla, Microsoft or whatever, but those jobs are increasingly tight. And so there needs to be looking at other markets. So this is the other market. The other market is lot of there's a ton of money going into AI startups. AI startups need tech talent. They need the best tech talent. And so this is the idea. The idea is my challenge is to see if I can develop a tool which helps connect those in need with those in need but
[7:47]With different needs. the the startups need the talent. The the international students need the job. So that's the idea. So I'll be basically blogging every day. This is a recruitment. So I also because this is a nonprofit, I going to do this directly. So if you want an OPT project and work on this one directly, I will supervise this one directly and you'll meet with me a couple times a week on this project. Uh but for this project um 80 days to stay what we're going to do is we're going to build a tool which connects startups with the best talent including international talent. So that starts
[8:33]With data. So the first thing is well where are we going to get this data? Where are we going to get this data about sort of every startup in the US? Well, turns out that if you raise money, that is you get investors, angel investors, VC investors, whatever, you have to file that, even if you're a private company, you have to file some basic information with the SEC, like who are you, what's your name, what's your address, how much money did you raise. This is different. You have to file a lot more of your public offering, but we don't care about that. We probably care
[9:05]About, do you have money? If you have money, you can hire. If you have a lot of money, you're looking tire. Later on, we'll start once we find the names of companies, start looking at the job boards and things of those companies. But the first step in any project is to get the data. And so, um, I just asked some of my best friends, Claude, Chpt, and Gemini, you know, where where can I get this data? you know, going to crunch that wouldn't allow me just to grab four million startups and you know that's not what they do. They're they're their thing. So where can I get this? It turns out that this
[9:48]Data is public. It's called SEC form D. So basically not every company does this but they are required to do this. And certainly if the company's big, a million, five million, which is mostly what we care about, 10 million, 15 million, then they're definitely going to file this form because they don't want to, you know, get shut down by the SEC after they've raised $100 million. The ones who tend not to file are the very small ones who just not aware that they have to because they don't have a big legal team telling you that you are required to fire file this form. Nevertheless, this form is very easy to
[10:22]Get the data. So basically we just download the data. So let's talk. So just the school gov. So basically it's an XML files and we just click download download download and then parse the XML. So that's tomorrow's task is to look at the data in more detail and um see what's there um what's in you know so what who are they but basically you know what it covers is you know name date you know industry etc etc which is really all the information we want we just want you know a you know a name we want where are they located we want how much money how
[11:02]Many investors who are in their investors Um, and then we're going to use some AI to then make some assumptions based on this. But they have a lot of money and few people they're hiring that kind of thing. But that's step two or three or four or whatever. So our first thing is just to it's just to get this data into not XML but other formats like TSV files or JSON files which is a trivial thing to do but that'll be day two. a day one which is actually quicker than I thought it would be is where can I find the name of sort of every startup
[11:41]In America and apparently if you're a startup in America has raised money you have to file with the SEC and because SEC is a public thing it's public data so we just download it and start looking at it so that'll be day two and day three or four if you're in my data classes this is a great project for you because it's about taking a data set and then enhancing it and enhancing and enhancing and enhancing and enhancing it. So we're going to infer for example do they currently um hire international students by who like who is in their company LinkedIn if they're all names like John Smith
[12:19]They probably haven't hired many international students but if not maybe they have but we'll talk about that later we'll talk about how to annotate data but this this is a surprisingly good start for day one. So basically it's fairly easy to get sort of the names, the addresses, the principal people, not the details, but the principal people of every startup in America. And so that's that's day one. So day one has been successful. Um yeah, and there's people who say they'll do that, but they don't really. It's super expensive. $99 a month is not for getting the data. $99 a month is for like typing in a few things and getting
[13:03]A couple thousand back in a Excel sheet. LinkedIn you have to scribe Angel, but this is this is this is what what we want. We have investor account. We have the fact that this is the government. People are not going to lie to the government and they might make mistakes but people are not going to willingly lie in a forum. That's problematic. Who are the founders? Um, all US companies must do this if they raise money. Well, if you're a US company who's raising money, you got to do this. Once you've raised money, maybe you haven't raised any money, then you don't have to do it. Uh, but if you're
[13:45]If you're actively seeking investors to invest in your company and then they invest something, even angel investors, you have to file this form. Of course, not everybody does, but we don't care really about everybody. We we care about those who at least a million in funding and probably five million in funding and more. So that's day one. Uh this is a project that you can join um you know to help you know my goal of creating this thing in the next 80 days. Okay, that's it. Take care.
More from 80 Days to Stay
10:53Building a Visa Sponsorship Database from SEC Data | 80 Days to Stay Project
7:31Why LinkedIn Failed Us: Building a Real Database for International Student Hiring (80 Days to Stay)
2:19LLMs Are Like Puppies: They'll Always Fetch, But Sometimes The Wrong Thing | 80 Days to Stay
4:15Building a AI-Powered Web Scraper to Help International Students Find Visa Sponsors | 80 Days 2 Stay
3:00From 500,000 URLs to Real Jobs: Inside the 80 Days to Stay Engine
2:08