He couldn't build the product he imagined. So he went back to school and learned how - then moved a billion dollars.
Gil Akos runs Astra, a company most people will never see and millions of dollars pass through every day. It is payments infrastructure - the plumbing that lets a software company push funds from one bank account to another in real time, instead of waiting days for the old rails to clear.
That is the present tense. Astra has processed more than a billion dollars in lifetime payments and, in the first week of 2025, doubled its annualized volume to over two billion. Akos talks about those numbers the way a careful engineer talks about a load-bearing test: useful, encouraging, and absolutely not the point. "We're building toward processing tens, even hundreds, of billions annually," he says. The milestone is a checkpoint, not a finish line.
What makes the story worth telling is how he got the credentials to build it. Akos is not a career fintech operator. He is an architect. He studied architecture at the University of Kansas, then went to Columbia's Graduate School of Architecture, Planning & Preservation for advanced architectural design. Somewhere in there, in his spare time, over a single semester, he taught himself five programming languages. People who knew him then knew him as the design-school student who kept wandering into code.
Then 2008 happened. The recession gutted the architecture industry, and Akos was laid off. The thing about being trained to design systems is that the instinct does not switch off when the building stops. He kept designing - just not buildings. He co-founded Mode Lab, a studio working at the intersection of computational design, automation, and AI, served as its technology co-founder and executive chairman, and taught at Pratt Institute. The architect was becoming a technologist in public, one project at a time.
The idea for Astra arrived before the ability to build it did. Akos wanted to help ordinary people make better decisions about their money - smart financial modeling, powered by deep learning, that would do the math people never have time to do. He knew the idea was good. He knew the product was needed. He did not yet have the skills to make it real.
Most people stop there. Akos enrolled in Udacity's Machine Learning Engineer Nanodegree and kept going - finishing roughly four and a half nanodegree programs across machine learning, artificial intelligence, deep learning foundations, and React. He treated a knowledge gap as a syllabus. The app he wanted to build became the app he could build.
"If you have an idea for an awesome new product, you should be able to sign up, integrate easily, and have funds flowing - without taking a whole bunch of risks."
- GIL AKOS, ON WHY ASTRA EXISTS
Ask a developer how long it takes to add payments to a product and you will hear a number that sounds like a prison sentence: twelve weeks, minimum. Contracts, compliance reviews, integration sprints, and the quiet terror of holding other people's money. Astra's pitch is blunt. Go live in about seven days.
That speed is the whole thesis. Akos built Astra as connective tissue between modern bank rails and the companies that want to use them - vertical SaaS platforms, marketplaces, payroll providers, anyone who needs to pay someone right now and not on Thursday. Visa Direct, RTP, FedNow, ACH: instead of forcing a customer to choose and integrate each one, Astra routes across them.
The use cases read like a tour of where instant money matters most. Merchant fast funding, so a small business gets sales proceeds the same day. Earned wage access, so a worker can reach pay they have already earned before payday. Instant disbursements, recurring transfers, card-to-account moves. The unglamorous machinery of getting the right amount to the right place at the right time.
Akos frames the company's growth as a partnership, not a transaction. "If they grow, we grow," he says of Astra's customers, "and these milestones are proof that this works." It is a tidy summary of a platform business, and also a tell about how he thinks. An architect does not get to be right while the building falls down. The structure either holds for everyone or it holds for no one.
Studies architecture at the University of Kansas, then advanced architectural design at Columbia GSAPP. Teaches himself five programming languages in one semester.
The recession hits the architecture industry. Akos is laid off - the first crack that turns a designer toward technology.
Co-founds Mode Lab, working in computational design, automation, and AI. Teaches at Pratt Institute. The architect goes full technologist.
Co-founds Astra with Sam Morgan to bring smart, automated money movement to consumers and developers.
Completes roughly four and a half Udacity nanodegrees - ML, AI, deep learning, React - to build the technology himself.
Astra raises a $10M Series A alongside a $30M credit facility, bringing total funding to about $40M.
Astra crosses $1 billion in lifetime payments processed and a $1 billion annual run rate.
In the first week of the year, Astra doubles annualized volume past $2 billion, aiming for another 5x growth year.
"He knew the idea was a good one. He knew the product was something the world needed. But he didn't have the skills yet to make it a reality."
- ON THE GAP AKOS CLOSED BY GOING BACK TO SCHOOL
He learned five programming languages in a single semester at Columbia - in his spare time, between architecture studios.
His pivot to tech started with a layoff. The 2008 recession ended his architecture career and quietly began another.
He earned four and a half Udacity nanodegrees to build Astra - a founder who treated a skills gap as homework.
He taught at Pratt Institute and ran Mode Lab before money movement ever entered the picture.
Astra was rejected by Y Combinator five times across YC and Startup School - and the team wrote about it openly.
He talks about a billion dollars processed as "deeply rewarding" - then immediately reframes it as motivation, not arrival.
Akos is explicit about where this goes. The aim is to make instant, automated money movement seamless and universally accessible - so that any developer with a good idea can wire up payments and watch funds flow, without taking on the risk that used to come with the territory.
He sees embedded payments adoption accelerating and is steering Astra into new verticals - healthcare, education, anywhere a payout needs to land faster than the old rails allow. He talks about folding AI deeper into payment operations to make them smarter and more secure, which is a fitting destination for a CEO who learned machine learning specifically so he could build this. The architect who once drew load paths through buildings now draws them through the financial system, and the structure is still going up.