Farnoud Kazemzadeh, BSc ’08, wended his way from an undergraduate “love for extragalactic astrophysics”, through a PhD in system design engineering, to become VP Engineering at Vital Bio – the second of the two startups he co-founded – where the techniques they are developing may one day allow your physician to replace an entire slate of tests at Labcorps with a 20-minute in-house test.
Oh, and this is after he spent a couple of years hoping (and training) to become an astronaut.
One of the most surprising aspects of Farnoud’s career, though, is how effortlessly natural (albeit awesome) he makes the path from studying freshman physics to performing medical tests seem.
His love of research was fostered early when working with Brian McNamara and Gretchen Harris as an undergraduate, where he and his colleagues “were literally measuring the weight of a supermassive black hole several million light years away.”
Farnoud took some time away from research during astronaut training, but after the astronaut corps didn’t pan out, he came back to UW to do graduate work. The problems he was interested in solving required tools that didn’t exist; as he put it, the telescopes were there, but the analytic tools weren’t. While working on making these telescope tools and worrying that the work wasn’t making as much real-world impact as he'd prefer, he and his advisor realized the same ideas could be used for biomedical imagers. This idea led him to add optical engineering, optical design, and photonics to his bag of tricks.
As a PhD student, Farnoud started combining machine learning with his existing work in a new codesign effort. Machine learning enables powerful techniques for replacing hardware components with software algorithms – for example, a physical lens made of glass can be replaced in software with a mathematical representation of a lens, saving 500g in the final object. Such tools are extremely useful for engineering hardware, where computational techniques let you make things smaller, faster, and better.
Speaking of his different perspective on how he sees software, Farnoud pointed out that
when most of us think about the use of software, “we look at our phone apps, we look at robots and AI; we don’t think so much about making simple, mundane things smarter.
But that’s exactly what he does.
Farnoud’s PhD project on biomedical imaging for tissue differentiation was more direct in its impact than his earlier work. He gave an example application of imaging certain types of brain tumors that cannot be directly seen because they do not take up dye. Without techniques like his, neurosurgeons have been forced to find these tumors by touch, probing the brain for differences in density, similar to how some people test the doneness of a steak by comparing its “give” to that of the muscle at the base of their thumb. The project eventually led to designing a device for skin cancer detection – a “digital biopsy” where you can diagnose cancerous/benign tumors without physically removing flesh.
At Vital, they’re now “decentralizing lab medicine” and telling apart different biomarkers in your body. “What we’re trying to do now is: most tests you would otherwise need to go to LabCorps to do... you can now do in a primary care physician's office with less than 1 mL of blood, running 70 tests in less than 20 minutes.” The work is much more complicated than the simpler problems of his graduate studies with an imager and an AI, requiring chemistry, fluidics biochemistry, biophysics and to find ways to “label” things so they’re easier for the analyzer to “see”.
Farnoud looked back at his days in physics and marveled in retrospect at the depth of the skills he had been taught and taken for granted. “In observational astronomy, there’s so much computation being done, it’s like this is obviously what you do, it’s table stakes. In engineering people spend years coming up with algorithms that [astrophysicists] don’t even think about, we just do it. Then you realize, wow, that algorithm in itself could have been a major publication.”
In a recent Zoom call, his former mentors Brian and Gretchen asked Farnoud what they’d taught him that he is still using.
His answer? The free body diagram.
It was a joke but what it meant was: you have to think about what are all the things acting on this object? You try to get to a root cause, a problem statement, and try to solve the problem considering everything that’s acting on it.
That analytical approach has served him well.
He appreciated, more generally, the mindset instilled by several years studying physics. “Just sit down and think about the problem and appreciate the problem. Let’s not just go out and look for solutions, any solutions, let’s think about what’s the best way to solve the problem and how to get the right answer.”
When asked whether he has any hobbies outside his work, Farnoud first mentions his two kids. He then casually mentions his work in the National Basketball League of Canada. (“I was never a very good basketball player!” Just a professional referee.) He also enjoys playing volleyball and does some woodworking these days, though he is careful to clarify that all he means by that is “nailing two pieces of wood together”.
It’s been a circuitous path, but even once Farnoud was on a path far from physics, he always remembered his undergrad department fondly. He was thrilled to hear his interviewer’s personal connection with Richard Epp. “Richard was one of the best teachers I’ve ever had, and best university professors I ever had. I always wanted to be like him. I learned most of how to think about physics from Richard.” And when working on the Waterloo campus, he reminisced, even while doing his graduate work in a different department,
Any time I opened the door to the physics building, any door, and smelled the building, it was just like coming home.
