Iris Mobility: Beating traffic in its own game with AI - The New Indian Express

Excerpts
You left India at the age of 17 to study computer engineering in the US. Looking back, what do you think that move taught you beyond academics?
Beyond academics, I learned how to live on my own. It was the first time I was away from my family. For the first 17 years of my life, I was always with my parents and had family around me. I learned to do a lot of things on my own — cooking, managing work, figuring out my schedule, keeping up with classes and friends.

Before founding Iris Mobility, you spent years working in computer vision and machine learning. At what point did you realise AI could play a meaningful role in public safety?
We’ve been working with artificial intelligence and machine learning since my junior year in college. Since then, I’ve been fascinated by how new models evolve every day and how we can use them to get better results, make better decisions, and do it quickly. When we were brainstorming ideas for what eventually became Iris Mobility, we focused on how to incorporate new technology to achieve the best possible results. We started working with a lot of machine learning models and crafted our solution from there.

Can you take us through what Iris Mobility is about and the kind of work the company does?
Iris Mobility is primarily a machine learning and public safety AI company. We work with organisations ranging from city governments, law enforcement, and traffic departments to commercial properties, construction firms, and school districts. We use existing infrastructure along with our own hardware, integrating it with our software to provide traffic analytics, safety analytics, pedestrian counts, near-miss crash detection, and insights into dangerous behaviour so action can be taken before accidents happen.

Iris Mobility focuses not just on crashes but also near-miss incidents. Why are near misses such an overlooked but important data point?
Near-miss crashes are overlooked very frequently because there’s usually no one there to validate what happened. For every crash that happens, there are around 20 near-miss incidents that have already occurred. If someone had identified them earlier, the accident might have been avoided. Near misses develop patterns. For example, at an intersection, there may be a recurring issue where vehicles taking a left turn are repeatedly at risk. It could be due to lane spacing, signal timing, or improper road markings. When we identify those patterns — situations where people narrowly avoid accidents by swerving or reacting suddenly — we analyse them and come up with solutions to prevent future accidents.

Was there a particular incident that inspired you to start Iris Mobility?
Yes, it was a personal incident. My co-founders and I were near the university one day, walking around and talking after classes, when we witnessed someone getting injured near a stop sign. Thankfully, the person was safe and taken to the hospital. But what stayed with us was the thought that if we had been there 30 seconds earlier, it could have been us. That moment made us want to work on a product that could reduce accidents. We sat down, explored different ways to solve the problem, and eventually arrived at Iris Mobility. Initially, we started with detecting stop-sign violations, and then gradually evolved and improved the platform into what it is today.

What makes Iris Mobility different from existing smart traffic or surveillance systems already being used globally?
What makes us different is the spectrum of services we provide. We don’t just focus on traffic monitoring or intersections. Using our core machine learning model, we work across sectors — from traffic organisations to construction firms.

For construction companies, for example, we provide productivity and safety metrics. Workers can be monitored to check whether they are wearing hard hats and safety vests and whether the site is safe overall. What truly differentiates us is how we approach every problem.
We employ cutting-edge technology, continuously update our systems, and bring a personal approach to solving these challenges.

What kind of response have officials given after seeing the insights generated by Iris Mobility?
They love it. Government officials understand that we’re trying to solve a critical problem. When we demonstrate how the product works and the kind of results it can deliver, they openly tell us that we are making their work easier. There are many workflows involved when different government departments communicate with each other. Our reports clearly outline the problem, proposed a solution, and the steps needed to implement that solution. That helps officials focus more on improving traffic flow and reducing accidents for residents.

Deploying AI in public infrastructure comes with concerns around trust and surveillance. How do you address fears related to privacy?
We make sure to stay updated with the latest cybersecurity regulations, and we work with a dedicated team that focuses on keeping data secure. At the same time, we are strong advocates of public privacy. We do not store information about cars, people, number plates, or vehicle types. Our machine learning model runs on the edge, which means that when something happens, only the count increases. For example, if a car passes through an intersection, the system simply records an additional count. We do not store pictures or personal information. The data we use is entirely numerical and meant only to generate solutions while protecting privacy.

Can you explain how the technology works in a real-world situation for a regular driver?
For example, imagine you take the same route every day and stop at an intersection where the opposite side has no traffic, but you still have to wait for the signal to change from 60 seconds to zero. When we work with traffic departments, we propose solutions involving signal pre-emption. If there’s heavy traffic moving northbound and no traffic moving southbound, there’s no reason for the opposite signal to stay green for a long time. Our technology adapts the signals dynamically so that the side with heavier traffic gets priority. Over longer corridors with multiple signals, turns, and exit ramps, this helps ease traffic flow, save fuel, and reduce road maintenance costs.

You mentioned plans to bring Iris Mobility to India as well. As someone from Hyderabad, what are your thoughts on traffic in the city and how can the technology help?
Hyderabad has grown tremendously over the last few years. There’s been a major increase in population, which naturally leads to more vehicles, traffic, and accidents. For example, areas like the Financial District and Durgam Cheruvu experience heavy traffic every day as people commute to offices, cafes, and other destinations. By analysing traffic data from existing cameras, we can identify alternate corridors for different times of the day. There may be one group travelling around 8.30 am or 9 am for work and another group travelling later in the day. We can propose alternative routes, work with the local police and traffic departments, and help create awareness among commuters. One of the biggest ways we believe we can help is by reducing accidents. There may be areas where wrong-side driving or jumping signals is common. Identifying the exact problem and the reason behind it gives us valuable insight into how those issues can be solved.

What does the roadmap ahead look like for Iris Mobility?
Our goal right now is to reduce the number of preventable accidents to zero. We’ve started with the city of Richardson, and we want to expand across Texas. Going forward, we also want to deploy in India — starting with Telangana, expanding to Andhra Pradesh, and eventually anywhere we can make an impact.