Dhruv Maheshwari Is Turning the Stop Sign Into a Feedback Loop

A stop sign is a remarkably optimistic piece of infrastructure. It has no lock, no gate and usually no witness. The city plants a red octagon at the edge of an intersection and asks a stranger in two tons of metal to make a good decision. Most do. Enough do not that a quiet residential corner can become a daily exercise in nerves.

Dhruv Maheshwari has spent years looking at that gap between instruction and action. At Obvio, the company he co-founded with Ali Rehan, the stop sign becomes the beginning of a feedback loop. A solar-powered camera watches the intersection. Computer vision running on the device distinguishes a full stop from a roll-through. Serious events can move to a human reviewer, then to local police. Ordinary footage stays local and is deleted on a short clock.

The machine is compact. The product is not. It includes legislation, a city contract, a roadside installation, public explanation, police judgment, driver education, privacy limits and a way to learn whether any of this changes behavior. Maheshwari's career makes more sense when viewed through that larger frame. He has been working toward systems like this since well before Obvio had a name.

01 / The first civic system

The education was outside the classroom

As a high school student in the Bay Area, Maheshwari started Einstein Explorations, a nonprofit that brought hands-on science lessons to under-resourced middle-school students. At the University of Pennsylvania, he enrolled in the Jerome Fisher Program in Management & Technology, the joint course of study that joins Wharton with engineering. He was drawn to the overlap among business, computer science and public service.

One summer he worked in West Philadelphia, spending three months with middle-school students, parents and teachers. It supplied the kind of product research that no dashboard can: daily contact with the people who live inside an institution's constraints. He later said those vivid memories kept pushing him to make social impact part of his career.

In 2014, Maheshwari and fellow Penn students started Hack4Impact. The model was practical. Nonprofits often needed software but lacked the money or technical staff to build it. Student developers wanted to use their skills on consequential problems. Hack4Impact made the match and organized the work. The first group had about 25 students; the network later expanded across universities.

“The vivid memories of students I worked with have pushed me to constantly want social impact work to be a part of my life and career.”Dhruv Maheshwari, reflecting on West Philadelphia

The pattern was already visible. Start with a public need. Bring technical people close to it. Build with the institution rather than around it. Measure success in usefulness, not in code shipped.

02 / Learning to see

From pixels to physical consequences

After Penn, Maheshwari worked at AltSchool and LinkedIn, then moved into augmented reality at Google. On ARCore, his work touched the computer-vision systems that help a phone understand planes and surfaces in the physical world. He also worked on privacy principles for augmented reality, an early encounter with a question that would become central at Obvio: what should a seeing machine be allowed to remember?

At Motive, formerly KeepTruckin, the connection between pixels and consequences became direct. Maheshwari led product work for computer-vision dashcams while Rehan led the AI and vision engineering. Their systems went into more than 200,000 commercial vehicles. The company reported that the product helped cut traffic incidents by more than a quarter.

Dashcams taught them what a model can recognize, how hardware survives a hard environment and how feedback can alter driving. The work also exposed a broader road-safety problem. Commercial fleets had begun adopting sophisticated systems, but a school crossing or neighborhood stop sign still depended on intermittent observation and the hope that a driver would cooperate.

03 / The city is part of the product

Code meets the permission layer

Maheshwari and Rehan did not begin with a clever camera and search for somewhere to place it. They read road-safety literature, attended conferences and stood at intersections. They spoke with parents, dog walkers, crossing guards and public officials. The same anxiety repeated: a recent near miss, a driver who treated the sign as optional, a crossing guard who felt exposed while doing the job.

The founders saw three established approaches to road safety. Education asks people to behave better. Engineering changes the street, often through expensive, slow construction. Enforcement creates consequences but is limited by staff and public concerns about discretion and escalation. Obvio's design tries to bind the three together. The bright roadside unit makes the system visible. The camera measures. Warnings or citations deter. New data shows whether the intervention works.

This is also where a familiar startup reflex runs into public space. Faster is not automatically better when software participates in law enforcement. The technical answer has to include limits. Obvio processes the continuous video feed on the device. Footage leaves when the system detects a suspected violation; ordinary footage is deleted after roughly 12 hours. A human checks the event, and a police officer retains the decision about whether to issue a citation.

Maheshwari put the challenge plainly: cities and their citizens need to trust the company. In this market, trust is not a mood created by marketing. It is a retention policy, a review screen, a public meeting and a decision to decline tempting uses of the data.

04 / The metric after the ticket

When success means less enforcement

Early Obvio programs in Maryland made the feedback loop visible. In Morningside and Colmar Manor, local officials described hundreds of daily violations before intervention. The cameras gave officers a way to review far more events than a patrol at one corner could catch. More important, Obvio and its city partners reported that stop-sign violations fell by half in the first eight weeks. Later results across school zones showed declines approaching 70 percent after four months.

That outcome creates a useful tension. Obvio supplies hardware without an upfront municipal charge and earns money through a share of citation revenue, with arrangements varying by local law. Yet the civic goal is to issue fewer citations because drivers stop violating. Maheshwari's case is that the model can work by focusing on egregious behavior across many locations, warning before punishing and staying responsive to the communities where cameras operate.

It is a harder story than “AI makes roads safe.” A camera cannot redesign a dangerous intersection. A model does not understand every emergency or human context. A citation can be wrong, which is why review matters. Public legitimacy can evaporate faster than a device can be installed. Obvio's proposition is narrower and more testable: persistent measurement plus visible, accountable deterrence can change a specific class of dangerous behavior.

“What is the right way to help people change their behavior, and then what is the underlying technology that can support that?”Maheshwari on designing from the outcome backward

05 / Scaling the whole system

A company built between interfaces

In June 2025, Obvio announced a $22 million Series A led by Bain Capital Ventures, with continued backing from Khosla Ventures and Pathlight Ventures. By then, the company was operating in five Maryland cities. The money was earmarked for national expansion, hiring and further product development.

Scale here will not look like a viral app curve. Every new community brings local law, road design, enforcement practice and political history. Maheshwari's job is partly to make that variability repeatable without pretending it does not exist. The company has passed laws in state houses, built camera hardware, trained computer-vision systems, staffed review and ticketing operations, and traveled from city to city to explain the program.

The latest work widens the lens. A driver-behavior study with Bicycle Colorado observed 50,000 vehicles at 196 intersections in 25 communities. Obvio has demonstrated distracted-driving detection at national road-safety gatherings and continued to participate in legislative discussions. The company is moving from one behavior at one corner toward a broader operating system for what happens on streets.

For Maheshwari, the route from West Philadelphia classrooms to computer vision at the curb is less crooked than it first appears. Hack4Impact taught him to place technologists inside a social need. Google made digital vision tangible and raised the stakes of privacy. Motive connected product decisions to driver behavior. Obvio combines the lessons in public, where the users include the driver, the officer, the crossing guard and the resident who never agreed to become data.

The red octagon still makes its optimistic request. Maheshwari is building the machinery that lets a city see what follows, respond in proportion and learn whether the request works better tomorrow. The ambition is not a street crowded with cameras and citations. It is a street where the feedback loop has done its job and the driver stops before anyone has to ask twice.