I've spent a career learning how.
I grew up in Shiraz taking things apart to understand them, and I never stopped. That instinct carried me from being the first woman to teach engineering at my university, at twenty-five, to a career building and leading AI in the places that punish a wrong answer: mining and welding floors, the systems that keep unsafe images off online shopping, the factories that quietly make the electronics in your pocket. Everywhere I went, the same truth held, the technology was never the hard part. Trust was.
I started Parseus to build the kind of AI that earns that trust from the start, for the person facing a hard decision and the enterprise running a thousand of them. The name carries the idea: Perseus, the Greek hero who won not by force but with the right tool used wisely, and Pars, the old name for Persia, a land defined by craft. Parseus is both, and so is the work: AI that does the heavy lifting, shows your options, and leaves the decision to you.
How I got here, the long way ▾
I grew up in a city of poets and wine, Shiraz, in the land of Cyrus the Great. As a kid, I was the sort of person who had to know how things were made. I wove Persian rugs, following a pattern knot by knot, where a single misread row pushes the whole motif out of line and there is no cheap way back, so you check every row as it forms rather than at the end. I embroidered by hand, where the thread and the daylight were finite and a mistake meant unpicking an afternoon of work, so I learned to plan the design around the cloth I actually had. I baked pastries, where the sequence and the timing are fixed and unforgiving: pull it too early or leave it too long and the batch is gone, with nothing to do but start over. And I played chess competitively, which taught me to plan several moves ahead and commit to a line knowing I would have to live inside the constraints it created.
None of it looked like engineering at the time. Looking back, the instinct was already there, the one that still drives my work: take a process apart until you understand it well enough to build something you can trust.
I loved all of it. But the thing that truly caught me was the computer. I was thirteen when I got my hands on a Commodore 64, and while everyone else wanted to play the games, I stayed up after the house went quiet, writing my own in BASIC. The crafts had taught me to follow a process someone else had set. Here was the opposite: I could write the process myself, line by line, and the machine would do exactly what I told it, no more and no less. If it broke, I had only my own logic to fix. That was the moment the work turned from making things to making the rules things run by, and I knew that was what I wanted to do.
That pull carried me from a computer engineering degree in Shiraz to a PhD in Waterloo and a postdoc in Vancouver. Along the way I built a solid scientific and engineering foundation that real work depends on, learned to invent new technology and push it past its limits, and came away with the habit I value most: never being done learning.
Before any of that took me abroad, my career began at home. Shiraz University offered me a faculty position in its school of computer science and engineering, having ranked as the top student across several years in both my bachelor's and master's.
I was the first woman appointed there in the years after the Iranian Islamic Revolution, when a woman's place in academic life was being newly questioned and constrained. I was twenty-five, standing in front of students who had never had a woman teach them at that level, and I learned only later that this became an inspiration for many young women to pursue their own ambitions regardless of the constraints around them.
The bridge into industry was a postdoctoral grant, one meant to push academic research into the real world.
I built everything myself and felt exactly how fragile an idea is before the world has tested it.
I turned research into scale, where something must work not once but a million times identically.
I owned systems where the margin for error nearly vanishes, where the bar isn't whether something works, but whether it holds under relentless load.
I built a vision system for one of the harshest places it can work, where a machine has to read molten metal through the same blinding arc that slowly takes a human welder's sight. Everything that makes it hard, the glare, the heat, the vibration, only shows up once you leave the lab and meet the reality.
I took on one of the most complex environments there is, close enough to the floor to watch an elegant idea collide with physical reality. I led the work that had to survive that collision.
I teach machines to drive by chasing what they've never seen, the rare, unimaginable moments that decide whether a car reacts in time. It's worse than finding a needle in a haystack: there, at least, you know what the needle looks like.
So I had lived on every side of that divide between the lab and the world, between startup and corporations, and each one taught me something the others couldn't.
The work itself was rarely tidy. I built computer vision systems for the mining and welding industries that had to see clearly through dust, welding arc light, constant vibration, and harsh weather. In mining, a single missed detection could jam a crusher, halting the operation and putting a life at risk to clear it. On the welding line, the system guided operators to lay a flawless weld straight through the blinding glare of the arc, solving a hard problem the industry couldn't escape: skilled welders are scarce, and their careers are cut short as they lose their eyesight by their early fifties. The impact was direct: lives saved, eyesight protected, and productivity raised sharply.
At Amazon, I worked on a few products, but two stand out. One was the chaos of trucks waiting to load and unload. Drivers stuck in hour-long lines that I helped turn into a few minutes, across hundreds of sites. It sounds like logistics, but it's really about a promise, getting people's orders to them on time. The other mattered to me more personally. I built the systems that catch unsafe and inappropriate images before they reach the public website. A miss could put harmful content in front of anyone browsing, including children, so the machine did the heavy work of catching what it could at scale, but on the calls that carried real consequences, a person made the final decision, never the model alone.
Later I took on an older world entirely, the factories that build the circuit boards inside nearly every device we use. I built that sprawling physical operation something like a nervous system, connecting and centralizing the data and knowledge scattered across its sites. The plant could finally anticipate problems instead of reacting to them, plan and schedule production on evidence instead of guesswork, and run with transparency and visibility at every level.
The technology was never the hard part. Trust was. That's the lesson every one of those projects taught me, over and over. A system can perform brilliantly in testing and still fail the moment it goes live, when no one can see how it reached its answer or hold it accountable. That isn't just useless. It's dangerous. What made these systems work was never how advanced they were. It was the discipline around them: their reasoning made visible, their limits clearly drawn, their performance measured against real outcomes, and a person kept in charge of the decisions that carry weight, someone who can see what the machine sees and still overrule it. The machine does the work, and you keep the judgment.
I started Parseus to solve hard problems the way I'd learned to: systematically, strategically, with the right technology for the job rather than the trendiest one. AI is a powerful part of that toolkit, but it's a means, not the point. The point is to build trust in from the start, not bolt it on after something breaks.
There's a story behind the name. It blends two ancient worlds in one. From Greece, Perseus, the hero who beat the impossible not with force but with the right tool, used wisely. From Persia, Pars, its old name, and a land defined by craft. Parseus is both: the courage to take on hard problems, the craft to solve them with care.
Everything I build at Parseus follows one rule: technology should hand people a clearer view and a faster path to a good decision, not quietly make the call for them, whether it's an app helping someone through a personal decision or infrastructure governing a fleet of enterprise agents. That isn't a marketing line. It's technology that respects the people who use it, does the heavy lifting, tells the truth about what it's doing, and leaves them more in control, not less.
I built Parseus to give people what they actually need, while the rest of the industry sells them what's fashionable in AI right now.