These student coders win IBM’s $200K top prize

During a trip to their village Yellapur in Uttara Kannada district in Karnataka last December, Hrishikesh Bhandari’s mother fell severely ill. Doctors back home, in Goa, explained to the family that the high level of arsenic in the village borewell water had triggered an acute immune response in her body.

Hrishikesh, a data science student at

IIT-Madras

, wanted to get to the root of the problem. He shared his concerns with four friends who he had met at an

Atal Tinkering Labs event

. One of them, Satyam Prakash, is also his batchmate at IIT-Madras. Then there is Manikanta Chavvakula, a business analytics student at Flame University, Jay Aherkar, a pre-university student, and

Sanket Marathe

, who is studying data science in Goa.

Together, they brainstormed. They found that almost 50% of the world’s population is still dependent on groundwater, and a third still have no access to safe drinking water. “We also realised that there is no mechanism to continuously check the quality of groundwater. We wanted to find a solution that keeps every stakeholder informed on water quality,” Hrishikesh says.

It was around this time that IBM held its global Call for Code programme, one of the largest “tech for good” initiatives. The five friends jumped right in. Over the next seven months, they built an AI-IoT platform called Saaf Water, which went on to win the competition’s top prize of $200,000.

The team developed a complete product suite consisting of a cellular-enabled hardware component, and a software platform that runs primarily on IBM Cloud and IBM

Watson

IoT services. While Hrishikesh is responsible for the overall project management and software work, Satyam takes care of the hardware component.

The hardware is a plug-andplay device fitted with sensors that check parameters like pH value of the water, total dissolved salts (TDS), turbidity and temperature. “Any local plumber can install it at the source of the water. It just needs power supply,” says Satyam. The microcontroller in the box relays data from the sensors to the cloud and the Watson IoT service, where it is processed. The box is designed to function even on 2G bandwidth.

“At the backend, our algorithms check past data of that particular water source, other ground water data sets, and then look for patterns. For example, algorithms will look for TDS, electrical conductivity, temperature, and other derived parameters, and can predict if there is a presence of heavy metals,” says Hrishikesh.

The solution does not replace testing. The device can periodically update the local administration, so that they know which water source to test.

A visual indicator is also built into the hardware. If the water quality is very bad, the device will display a red signal. “That allows even a user without a mobile connection or access to the Saaf Water dashboard to know about the water quality,” says Satyam.

The team has hosted Saaf Water on open source platform GitHub and is looking to build a community around it.