Somewhere in a warehouse, an old laptop is having an existential crisis. Its battery is tired, its keyboard has opinions, and its corporate owner has moved on. The machine may still have years of work left. Yet before it can reach a student, a family, or a secondhand buyer, somebody must open it, identify it, test it, erase it, diagnose it, reinstall its software, photograph it, and package it. Multiply that minor administrative epic by a few million and reuse begins to look less like a moral choice than a labor problem.
Rupesh Jeyaram likes this sort of problem: important, physical, repetitive, and hiding in plain sight. In a Brooklyn workshop, his company Revise Robotics builds automated systems that can pick up mixed laptops and shepherd them toward another useful life. The robot arm is the visible bit. Behind it sit computer vision, software agents, custom hardware, inventory systems, and the unromantic patience required to make a cable meet the correct port over and over again.
The work brings together two ambitions Jeyaram has carried since high school. In an Earth Day reflection in 2026, he wrote that an environmental science class made him fall for the natural world and the work of protecting it. At roughly the same time, he discovered that software could change more than pixels on a screen. It felt, he said, like both a superpower and a responsibility. Ten years later, he could stand in the same room as robots refurbishing laptops for people who needed them.
“Software, sustainability, and impact, all in the same room.”Rupesh Jeyaram, reflecting on a decade of climate workChapter I
The climate work began in the code
At Caltech, where he earned a computer science degree in 2020, Jeyaram learned how environmental ambition could become an engineering assignment. As a scholar with the Schmidt Academy for Software Engineering, he worked with atmospheric scientists studying greenhouse gases and solar-induced fluorescence. Their research depended on a thicket of legacy tools written in Fortran and C++, tied together with Python. The academy's task was to make that software easier to read, use, and improve.
This was climate work without a wind turbine or solar panel in sight. Its material was code maintenance, the sort of labor that rarely makes a poster but determines whether a research group can move quickly. One methane-detection paper later thanked Jeyaram for helping with open-source radiative-transfer software and making its computations run faster. The episode offered an early version of a principle that would reappear at Revise: a worthy outcome can be blocked by an awkward workflow.
Caltech also supplied evidence that his attention was not reserved for machines. He tutored high-school and community-college students in mathematics and science through the Caltech Y's Rise Program, then served on its advisory board. One student, Ellis Sentoso, initially worried that a Caltech tutor would judge what she did not know. Jeyaram spoke about his own struggles with data science and coding, and the worry loosened. He later helped her through more courses at Berkeley and turned up at her commencement when she invited him.
He also led monthly groups of students who cooked dinner at a Pasadena adult center, and he served as president of his undergraduate house. At graduation, Caltech gave him the Doris Everhart Service Award for care and concern shown through campus and community work. The useful detail here is not that an engineer volunteered. It is that teaching, access, and technical confidence were already bundled together before circular electronics gave them a commercial form.
First, teach a robot to see a roof
Jeyaram went to Wharton with the explicit goal of starting a company. He focused his MBA on entrepreneurship and Business, Energy, Environment and Sustainability, while working at another young robotics company. Renovate Robotics was building a machine for an unusually unforgiving workplace: a roof. As computer vision lead, he developed the perception system for a robot designed to install roofing materials while keeping people away from hazardous work.
Screens are polite. Roofs have glare, shadows, uneven surfaces, loose material, changing weather, and gravity's full attention. Computer vision in that environment must become more than an impressive model. It has to remain useful when light shifts and the physical system behaves imperfectly. The experience put Jeyaram close to the discipline of embodied software - code whose mistakes have weight, distance, and consequences.
In May 2024, the same week he finished at Wharton, Jeyaram announced that he had also completed his last day at Renovate. He thanked friends and family, wrote that it took a village, and hinted that he was cooking up something in climate technology. The recipe turned out to include old computers, robot grippers, and Antonio Monreal, another Caltech graduate with a personal connection to refurbishment.
Monreal's first laptop was a half-broken machine recovered by a cousin who refurbished devices for sale in Mexico. It helped him learn English, study calculus online, apply for work, and eventually reach Caltech. Jeyaram brought the climate-software and robotic-vision thread. Together, they found a problem where access and environmental value could share the same production line.
The second life has an assembly line
Revise Robotics was founded in 2024 around a blunt observation. Americans discard more than 30 million computers each year, even though many retain resale value. The devices that make it back into circulation must pass through a sequence of slow tasks. A recycler or IT asset-disposition company receives mixed brands, models, conditions, and operating systems. Each one demands attention before anyone can trust it with a new owner.
The company's first system is intended to automate that route. A robot assesses a laptop, wipes its drive, runs tests, photographs the result, and prepares the device for resale. The phrase “mixed laptops” contains most of the engineering mischief. Unlike a car factory, where each object arrives on schedule and in a known shape, a refurbishment line inherits years of product design decisions. Ports migrate. Hinges resist. Menus vary. A BIOS can turn a simple key press into a diplomatic negotiation.
Revise's public demonstrations make the challenge pleasantly concrete. At Y Combinator's 2025 Demo Day, the system tested different laptops, plugged in cables, moved through software menus, and reinstalled Windows. Jeyaram's caption was less interested in robot theater than in the fate of the objects: billions of laptops, phones, and tablets, he wrote, deserve a second life.
The company won Best Pitch at the 2024 Pennovation Accelerator, joined Y Combinator's Winter 2025 batch, completed early pilots, and disclosed roughly $3.6 million in new equity financing in a May 2025 federal filing. Money is not the destination, but in hardware it buys iterations: more devices, more edge cases, more hours discovering that the port was two millimeters left of where the model expected it.
“There are billions of devices out there that deserve a second life.”Rupesh Jeyaram after Revise's Y Combinator Demo DayChapter IV
A robot arm, with consequences
The most revealing deployment may be the partnership with Computers 4 People, a nonprofit that provides refurbished devices to people who need them. In early 2026, the nonprofit said it was processing about 500 laptops a month with Revise. The machines were not being prepared for a speculative warehouse. They were going toward jobs, education, and everyday participation in a world that assumes everybody has a computer.
This is where Jeyaram's old threads tighten. The tutor who helped a student feel less intimidated by coding is now trying to make more computers available. The climate-software engineer who removed friction from atmospheric research is removing friction from reuse. The computer-vision lead who taught a robot to operate on roofs is teaching another to cope with the peculiar geography of keyboards and ports.
His public voice retains a useful bit of mischief. When the team took a robot to Las Vegas for an industry event, Jeyaram posted: “What I learned about enterprise sales from shrink wrapping our robot… Just kidding.” It is a compact portrait of hardware entrepreneurship. You may spend the morning discussing markets and the afternoon making sure several hundred pounds of machinery survive freight.
By summer 2026, Revise was recruiting people with “deep, weird strength” for problems such as vision under hostile lighting, laptop manipulation, BIOS navigation, and cable insertion. Jeyaram was also taking the system into industry conversations, including a presentation in Frankfurt on practical AI in circular electronics. The company's horizon extends beyond a single robot cell. Its stated target is the wider economics of electronics refurbishment, beginning with laptops because they are abundant, valuable, inconsistent, and stubborn enough to teach a young robot some manners.
The ordinary future
Robotics is often sold through spectacle: a backflip, a dance, a humanoid silhouette approaching the camera. Jeyaram's wager is quieter. He has chosen the workbench, where success looks like an uneventful afternoon. A laptop arrives. The robot recognizes it, opens it, plugs it in, navigates its software, records what is wrong, and moves it toward a new user. Nobody applauds. The device simply avoids the shredder.
That ordinariness is the charm. Circularity requires fewer declarations and more competent handoffs. Every material that remains useful must pass from one owner, process, or form to the next. When a handoff costs too much, the loop breaks. Revise Robotics is trying to lower the price of one particularly fiddly handoff, one that happens to contain both environmental value and access to the digital world.
A decade after his environmental science class, Jeyaram can measure the distance traveled in an oddly satisfying unit: refurbished laptops. Each one is a small argument that software can touch the physical world without merely making more of it. The robot may be the new machine in the room. Its purpose is to keep the old machines useful.