It starts with a blip

Israel brims with technology talent. But with the tech scene booming – especially for AI-focused startups – there’s also huge demand. And while tech people have honed their skills in many different areas, including fintech, ad tech, cyber and all the rest, the maritime domain usually isn’t one of them.

And that leads to the most common question I get asked by candidates: what is it that you guys actually do? Specifically, they want to know what the day-to-day of R&D work here involves, and how we apply our technology to help decision-makers to protect their interests and outsmart their rivals or adversaries in the maritime domain.

I break the answer down into three parts:

It starts with a blip

A “blip” is the term we use to represent a ship’s location at a specific point in time. Around 100 million new blips are added to our system every day. As part of this process, we “clean” the data to get rid of noise and inconsistencies. This allows us to put all the ship blips on a map to provide us with an accurate picture of the route a ship is sailing, or sailed in the past. From here, we can derive the ports visited, the speed traveled and any other variables shedding light on its journey and operations.

Screenshot from the Windward system

 

Behavioral features

Once we’ve created the travel path of a ship, we combine it with other information, including weather data, nautical charts, and port state control data, to create static and dynamic behavioral features. Static features include things like the age or size of a vessel. Dynamic features might include the time a ship spent traveling at dangerous speeds and/or unsafe waters, if the vessel carried out any ship-to-ship operations and/or carried out dark activities.

These features are then fed into our machine learning algorithms.

Modeling

The next step is for our data scientists to build predictive models based on these features. These allow our clients and partners to assess the likelihood of a vessel being involved in smuggling operations or other illicit activities; or having an accident in the year ahead; or being in breach of sanctions.

 

The infrastructure that enables all of this is based on Java and Python applications built by our Backend Developers for data-processing using Spark clusters; the data is kept in Mongo, Cassandra and ElasticSearch databases; caching is done via Redis. The results are presented to our customers via our user Interface, which is built by our Full-Stack Developers using React/Angular, and based on Java/NodeJS microservices infrastructure.

The process may sound simple, but it’s anything but. Only by employing the best and brightest technical minds, alongside super-skilled professionals, can we create a new layer of insight to support maritime decisions wherever and whenever they’re needed. It’s the only way to make an impact.

Alex Milman is a Senior Backend Developer at Windward