“I think we’re only just getting started with analytics. But finally the evolution is happening and the conversation about data science has become mainstream.”
Prasad Kothari has been in data science for almost 15 years, in which time he’s worked with the latest technologies, explored multiple industries and published two books. Here, he talks about some of the highlights of his career, from helping the US government battle addiction, to using natural language processing for effective contact centre management in his current role at The Smart Cube.
Making the numbers work
In his own words, our vice president of analytics grew up as a “maths nerd”. Prasad’s love and natural affinity for numbers was apparent from an early age, but what wasn’t immediately clear was how it would become useful.
By the time Prasad had graduated from university and secured his first job as an analyst in 2005, the data science industry was still relatively nascent. But its potential was clear.
“I saw all these different problems being solved with mathematical modelling, computer sciences, NLP and deep learning, and I was immediately interested,” he says. “One of my first projects was performing marketing optimisation on CRM data of a US-based telecom client, using NLP, deduplication and advanced statistical modelling to understand consumer buying and spending behaviours. The notion of solving business problems with the power of mathematics and machine learning was fascinating. So, I went to do a Master’s degree in Arizona to learn more.”
The science of addiction
Throughout a varied career, Prasad has applied his skillset to a wide range of use cases in a number of industries. One of his most memorable projects was the work he did with the US government and National Institute of Health (NIH) to help battle addiction.
“I came in as a statistical consultant to help develop treatments for drug and alcohol dependency,” he says. “We were conducting clinical trials where patients were undergoing multiple treatments for multiple conditions, and sometimes there were adverse effects. I built AI and mathematical models to help improve our understanding of these side-effects and limit their impact, while improving treatment efficacy.”
As addiction is both a physical and mental affliction, the use of data science wasn’t solely restricted to identifying the best medicines.
“One of my other research projects at NIH was similar to Match.com,” he says. “Except we were trying to match patients with the right psychotherapists. Trust is such an important factor in that kind of treatment, so we were using analytics to try and give everyone the best chance of recovery.”
This work became the catalyst for various policies and strategies in the US healthcare system, and the research is still being funded to this day. For Prasad, it was an experience he won’t quickly forget.
“Seeing data science impacting so many lives in a positive way was incredible, so I’m still really proud of those projects.”
A different challenge every day
His work with The Smart Cube has seen him tackle everything from extracting data from images to improve retail efficiency, to using sentiment analysis to read the facial expressions of movie-goers.
“I’ve spent a lot of time shuffling between domains and sub-domains,” he says. “But the fun part of that is I get to talk to so many different clients, see so many different perspectives and understand so many different business problems. It’s impossible to get bored.”
“We’re only just getting started…”
As the analytics industry is incredibly fast-moving, Prasad dedicates a lot of his time to keeping up with the latest trends.
“I try and read at least a book a week, normally on subjects like AI and deep learning,” he says.
Not just an avid reader, he’s also written two books of his own and published multiple research papers, accumulating over 300 citations in publications like the European Journal of Business Ethics.
With all that reading and research, and 15 years’ experience under his belt, there may be no one better placed to talk about where the industry is heading next.
“We’re only just getting started with analytics and AI,” he says. “But finally the evolution is happening and the conversation about data science has become mainstream. I think the next big thing we’ll see is real investment in computer vision and conversational AI.”
Computer vision is the process of extracting data from digital images and videos. The use cases here are widespread: from analysing MRI scans to identify tumours in their earliest stages or using Instagram pictures to spot early indicators of depression, to monitoring the facial expressions of consumers as they make purchase decisions in-store. It’s a relatively new area which started gaining traction in the last four or five years, but in a fast-moving industry, emerging technology can become the standard in the blink of an eye.
Similarly, the adoption of chatbots has seen a steady increase across a range of industries – from insurance companies trying to increase customer satisfaction to e-commerce brands working to improve Net Promoter Scores (NPS) through better use of conversational AI.
A quantum leap
Long term, Prasad thinks the industry’s biggest shift will come from superior quality data collection and validation through technologies like blockchain and the emergence of quantum computing.
“We’re probably about six or seven years away from seeing that,” he says. “But companies are investing tens of millions of dollars as we speak.”
Quantum computing would allow data scientists to not only have more computational power but also better algorithms, like quantum neural networks using entanglement principles. This will make it far easier to handle complex data with complex inter-relationships, and the implications could be huge.
“Think about life sciences,” Prasad says. “Right now, patients receive a standardised or semi-personalised treatment for cancer. But everyone has a different genetic makeup, and everyone is exposed to different environments, so each treatment works better for some than others.
“With superior quality data, real world evidence and quantum computing, we could provide truly personalised treatments for every patient. And drug discovery optimisation could be taken to a whole new level using quantum neural networks.”
“Taking an industry such as e-commerce, an advanced recommendation engine powered by hyperbolic deep learning and Poincare geometry for feature extraction, coupled with quantum neural networks and reinforcement learning, could deliver a next level of hyper-personalisation.”
This is the kind of development that could break the mould of data science as we know it. And at the rate the industry is moving, it might not be too far away.
Want to meet more of The Smart Cube’s data scientists? Keep an eye out on our blog over the coming weeks and months!