The dyno asks you to commit before you know. You crouch, you throw yourself upward, both hands leave the wall, and for a moment you hang in the air with no hold at all, aimed at a grip you are fairly sure you cannot reach. Saurab Dhir spent longer on that one move than on any named route, at Progression Bouldering in Downtown Vancouver, weeks of trial and failure and recalibration before it finally went. As a beginner it frightened him, which is the point. A dyno gives you no way to test the outcome first. You leave the wall or you don’t. His work is the same wager in slower motion. Dhir works at Mecka AI, where his job, in a sentence, is deciding which recordings of human work are real. Contributors around the world film themselves doing physical tasks, robots will eventually learn from the footage, and between the two sits a person who built the machines that judge authenticity, and who therefore spends his days in close contact with the many ways people fake things. It is a strange job to explain at a dinner and a fitting one to hold now, in a decade when synthetic text and images and video have made “is this real” a daily question for everyone and a rigorous one for almost no one. Dhir is among the few paid to answer it with instruments instead of instinct.
The temperament came before the job. Dhir grew up in a quiet neighborhood on the edge of Delhi, the son of a father who ran a manufacturing business and a mother who kept the house, and he was the kind of boy who could not leave a machine closed. He took an RC helicopter apart and never got it back together, which taught him more about his stubbornness than about rotors. Later bricked a phone trying to jailbreak it. For his fourteenth birthday Dhir’s father gave him the job of building his own gaming computer, not the computer itself: the sourcing, the research for value, the haggling, the buying, a boy sent to do an adult’s diligence. Video games pulled him toward engineering and never let go. Both parents shaped how he thinks about honest work. His mother taught him what empathy in work looks like. His father taught the harder thing: Dhir watched him reach decisions where the dishonest path was plainly easier and take the right one anyway, living by a single motto, that honest work comes at a cost far cheaper than dishonest work. The lesson underneath was to put others first and to care especially when caring is inconvenient. It is no accident that the son grew up to build machines whose only job is to tell whether a thing is what it claims to be.
What pulled him to Simon Fraser was an image. He had looked at photos of the campus online and thought, that is the dream, that is what carries me somewhere bigger. The arrival undercut the dream fast. His first month in Canada was, he says, the worst stretch of his life: he had never lived away from his parents or run a household, and his first night he slept on a bare mattress on the floor under a blanket borrowed from the landlord, in a room where he could watch spiders move on the wall. The night he remembers most is the one he called Rohan Parmar just to get out of the room, and the two of them walked the block until dawn talking about their ambitions. He moved onto campus soon after, then into his own place on Burnaby Mountain. That loneliness became part of the discipline. It taught him that the hard parts of life are not solved by cleverness alone but by showing up, by continuing, by making the small unglamorous choices that no one celebrates. In the cold of that first Canadian winter he learned to cook for himself, to budget, to sit with his own thoughts when there was no one else to talk to. By the time he found the bouldering gym, he had already started to understand that fear and commitment are not opposites but partners: the fear tells you the jump is real, and commitment is the only answer worth giving.
His colleagues meet the temperament as rigor; the recordings meet it as suspicion. Dhir’s systems have learned the physics of dishonesty, that a fake task leaves the world suspiciously unchanged, and he has learned the harder human corollary, that the line between a cheat and a perfectionist can be a few frames of hesitation. He talks about contributors the way a fair referee talks about players, watching for the dive but slow to call it without proof. The work has made him exact about a distinction most people blur: authenticity is a property of the world, and proving it is a discipline, with instruments and error bars and costs. There are people who submit recordings of tasks they never truly performed, and his job is to catch them without cynicism and without naivete. He has learned to read the micro-signatures of real effort: the way a hand adjusts grip after a slip, the subtle shift of weight before a lift, the breath that catches at the moment of strain. These are things a scripted performance rarely captures, because they are not planned; they are lived. But he also knows that a perfect recording can be a lie, and a messy one can be the truest thing in the world. So he builds systems that are patient enough to hold the ambiguity and rigorous enough to demand evidence. He says the goal is not to punish people but to protect the value of honest work, so that the robots learning from human movement are learning from something real.
Away from the screen, two threads. He picked up a guitar at thirteen, and music has stayed central since, Metallica and Lamb of God and Megadeth and Iron Maiden alongside softer, newer things like The Marías and Laufey. The other thread is chai. For about four years in the same apartment, two of them shared with Mikhail Sinitcyn, his Docula co-founder, the morning ran on the same ritual: grinding spices, boiling water, making fresh chai, and handing a cup to Misha every day as a small act of gratitude, for family and for friends. That is the detail he wants remembered. It is the kind of detail that does not appear on a resume or in a performance review, but it says more about who he is than any title could. The chai ritual was not efficient, not scalable, not supported by any product metric. It was simply a way of saying, without words, that the people around him mattered, that his days were not only about building systems and catching cheats but about tending to the small invisible bonds that make a life worth living. When he talks about it, his voice changes, softens, as if he is holding something fragile and dear.
The pairing is less odd than it looks. Metal is exact craft that sounds like chaos from outside; a dyno is honest failure repeated until one clean commitment lands. Neither lets you fake the hard part, which is the standard he now keeps on machines’ behalf. There is a version of this decade in which the people who certify the real become its quiet central profession, the way auditors came out of the joint-stock scandals and food inspectors out of the packing houses. Every age of cheap fabrication eventually mints the people who vouch for the genuine article. Dhir arrived early, almost by accident, following document mess to video mess to the problem underneath both. And the job, in the end, resembles the move he practiced longest. You gather the evidence you can, and then the moment comes to commit past certainty, both hands off the wall, trusting you have judged the reach right. You leave the wall or you don’t. That is the honest work he is doing, and at its core it is not about suspicion at all. It is about care: care for the truth, care for the people who do real work, care for the machines that will one day learn from them, and care for the small human rituals, like a cup of chai handed to a friend, that remind us why truth matters in the first place.