More than 1,50,000 people are killed in road collisions each year in India, which amounts to 400 deaths per day. If one were to ask if deadly road collisions can be avoided with technology, and the answer is yes. One such company is Zendrive which offers its technology solutions to make driving safer for users and help insurers save money in the process. The San Francisco and Bengaluru-based startup recently closed its Series B funding round at $37 million, from XL Innovate and Hearst Ventures and previous investors including ACME Capital, BMW iVentures, NYCA, SignalFire, and others.
According to Frost & Sullivan, Zendrive’s predictive risk model is six times more accurate than the industry average. This is because the volume of data the company owns is huge, and increases by 11 billion miles worth of data each month. The company’s risk model has also been trained extensively using advanced machine learning algorithms. Analytics India Magazine connected with Pankaj Rasbood, co-founder and VP Engineering Data at Zendrive to dig deeper about the company, its work culture and innovation.
What is Zendrive trying to accomplish in the road safety area?
Pankaj Risbood: This is a company with a mission to make roads safer. We try to understand how people drive their vehicles and using the insights that we gather from the behaviour analysis, we are able to do a lot of things. We not only help people to become safer drivers but also understand the kind of risks they have while driving. Then, use those insights for insurance premium, for predicting losses. We can also detect collisions when they happen all of this we determine using mobile phone data without any hardware. So, make the deployment easy for our customers.
When it comes to collecting data for making travel easy and convenient for users, what kind of trends do you see in that market?
Pankaj Risbood: It’s really about human behaviour at the end of the day. All of our culprits of driving distracted and this is a problem which has come up just in the last 10-15 years or so. Today, 25% of all road fatalities are caused by people being distracted on their phones while driving. So, being able to understand what is at the root of that, we can educate people to be safer on the road. It’s something that we find really important. It’s a difficult and scary moment for anybody when they are in a collision on the road. Being able to able to detect that autonomously and get them to help if they need it and saving lives in the process is something that we are proud of at Zendrive.
Could you tell us about Zendrive’s technology solutions? How are you leveraging AI to make driving safer?
Pankaj Risbood: The entire premise of Zendrive is about analytics, machine learning, artificial intelligence. We are using the data that is available from smartphones through various sensors. But that’s just data. All the intelligence on top of that which is to understand what patterns in data are telling us that somebody is using their phone, what patterns are telling us that somebody is driving well above the speed limit, what patterns are telling us that somebody is breaking the rules, those are the kinds of intelligence and insights that we derive out of sensor data. We do a lot of sensor processing, we build a lot of classification and regression algorithms for building driving behaviour insights.
We are also using these insights to generate scores and profiles for the drivers. That involves a lot of statistics, a lot of machine learning again to understand who is risky. That risk needs to be priced in insurance pricing, factored into the claims and so on. We use AI/ML pretty much throughout the entire company and all of our based off it.
Tell us what new is happening at Zendrive in terms of innovation.
Pankaj Risbood: We are always investing in exciting technologies because ultimately it’s a tech product that is serving a broad segment of the market. Just last month, we announced the launch of a new product feature called full stop whereby we are automatically detecting if an individual is not stopping at a stop sign but running over it. Now, you can imagine running through a stop sign is quite dangerous because cross-traffic can hit you and cause collisions. We are able to detect those things autonomously and automatically. Doing things like this requires being able to process data at scale.
We have petabytes of driving at this point in time. This is geo-spatial and temporal data at the same time. So, processing the data, being able to derive insights is something we are doing at scale. Deep Learning, the newer advances in neural networks is something we are exploring. We do quite a lot of CNN and LSTM, and those kinds of things given the temporal nature of our data. We are also investing in newer technologies which are privacy-preserving- things like Federated Learning and so on. We see those as the future where the world will eventually gravitate towards.
What kind of skills are you looking for your hiring process?
Pankaj Risbood: We have a 20-people data science team right now, which is growing and expanding all the time. We look for people with core statistics and machine learning background, people who have done either a Master’s or PhD in Deep Learning, statistics, artificial intelligence, and similar domains. We certainly look for people with strong coding abilities because ultimately it’s a product, it’s not research. So, being able to deliver the product is important going through the rigours of production deployment. And above everything, we look for people who are curious. Solving a problem like road safety is not easy. It’s one of those problems which has not been explored much. So, we look for people who have a lot of curiosity to go deeper and understand human behaviour so we can model that and use that in the solutions that we are providing. Business is really important but at the end of the day, there is a social component to it which is- hey we want to save lives on the road. People who are motivated and excited by that kind of mission is something we want to have.
Tell us about the learning and training programs inside Zendrive.
Pankaj Risbood: In the technology industry now, there is no point when you can say I know everything. There is always new stuff coming in. Even if you know a lot, there is always a need to learn and re-learn new things. Even the senior staff at Zendrive — the principal data scientists are always looking to learn about new things. At the same time, the engineers and data scientists we hire fresh out of college, we take them through a rigorous program of coaching and training in the first year, making sure they understand stuff so that they learn two things — one sort of the tricks of the trade what skills are required but more importantly how to learn in the first place because our education system doesn’t really teach you that. We think about a career as continuous learning and that’s something we try to imbibe into everyone.