Anmol Mohan

[box type=”shadow” align=”” class=”” width=””]

Currently based in Singapore, Anmol is the Chief Executive Officer of Tuple Technologies. Anmol brings to the table 8+ years of experience in Data Mining, Pattern Recognition, and Machine Learning. Throughout his career, Anmol has helped various clients across US, Europe, India and South-East Asia with their requirements in business, behavioural and marketing analytics.

Having graduated in Mechanical Engineering from Motilal Nehru National Institute of Technology, Allahabad, Anmol decided to make a career in a field of his choice where he can utilize his natural enthusiasm towards Logic and Mathematics. He took up an analytical position in consulting firm ZS Associates to begin his career but quickly developed an itch to create something on his own. To gain experience, he worked for several startups and numerous clients in India. Before serving as Head of Data Analytics at Paktor, Anmol was instrumental in setting up Data Mining team and processes from scratch in Metro AG, a Global Fortune 5 00 company.

Apart from his love of solving puzzles, Anmol spends his spare time playing and composing music. He plays seven different musical instruments and maintains a blog where he writes about social evolution. Anmol wants to use his knowledge to create technologies that bring advances in Data Science to the masses and help in the betterment of society as a whole.[/box]

Interview: Anmol Mohan with ComingTechs

[box type=”shadow” align=”” class=”” width=””]

Newsdesk: You have worked in many areas of data analytics, so what are the main applications you have worked with?

Anmol: I have worked extensively on Customer Behavioural Analytics. For any business, the customer is the most important factor. Most successful businesses keep a sharp eye on their customer preferences through their data.

In a fast-moving world, customer choices change very quickly, impact on a daily basis not only by their needs, but by a lot of external factors like Politics, Weather, Markets, and Technology.

There were simpler times when business owners had the luxury of knowing every customer personally. In the current scenario, successful businesses let data speak on behalf of their customers.

The intelligence that you apply on data lets you understand what the patterns emerging from data mean in terms of actual customer behaviour. Once you have grasp of the same, you can use these insights to optimize your Sales, Marketing, Supply Chain, Inventory, and Finance. I’ve had the privilege to work on most of these areas during my time with Metro, Paktor and now Tuple.[/box]

[box type=”shadow” align=”” class=”” width=””]

Newsdesk: How can your engine predict the future behaviour of individual customers?

Anmol: Generally, most prediction algorithms follow the same process. For example – You want to know which existing customers are likely to leave your brand in the next 10 days. The algorithm is trained to look back 10 days in the data. This is to find out which customers have left the brand in these last 10 days. Once you have this classification, the algorithm finds correlation between the behaviour shown by customers who had left prior to these last 10 days and the actions they took before leaving the brand. This is called a model.

Once you have this model, you can project it on the behaviour shown by other customers recently. This will help you predict which customers are likely to leave since they are likely to show similar behaviour.

No model is 100% accurate as they are mere statistical approximations. However, with proper knowledge of technicalities, iterations with algorithm selection and parameter tuning, one can achieve a relatively high accuracy of prediction. Currently, our production models have an accuracy of between 86% to 98%.

The unique thing about our engine is that we have automated the procedure of data cleaning, variable & algorithm selection as well as parameter tuning. Hence our engine can churn out the models with much higher accuracy and is almost 125X faster than the traditional methods of doing predictive modelling.[/box]

[box type=”shadow” align=”” class=”” width=””]

Newsdesk: Can you explain in detail and how it can be helpful?

Anmol: All the decision making that businesses do is actually nothing but forecasting. As heads of our departments, we will want to make choices that can bring maximum benefit to our organization. This can be in terms of monetary benefit or may be in the form of customer satisfaction. But all we are trying to do is to predict future outcomes based on current decisions and actions taken. We have done this sometimes through experience alone, sometimes through data alone, and sometimes through both.

Getting highly accurate predictions about customer behaviour can help a lot. Not only can it give you useful information to help you make informed decisions, it can potentially save a lot of money spent rectifying decisions that were made without.

For example – our engine optimizes Sales strategies by telling the sales reps to go after leads which are most likely to convert. As such, some of our clients have seen up to 56.3% increase in yearly profits by predicting which customers are likely to leave and proactively persuading them to stay. It is a win-win situation as customers get what they want with personalised care and brands profit more by keeping their customers happy.

Companies have limited time and resources. We have a solution that can help them utilize their limited resources for maximum gain.[/box]

[box type=”shadow” align=”” class=”” width=””]

Newsdesk: Any use case for Healthcare using your technology?

Anmol: Yes, of course. The technology behind our algorithms can be very easily be applied to healthcare projects. I have seen many startups using these algorithms to detect cancers from patient scans. Or predicting which patients are likely to develop heart conditions in future and taking remedial actions beforehand.

These are the same algorithms used in Self-Driving Cars, your daily voice assistants like Siri and in detecting breast cancer from clinical imagery. I am very glad that we at Tuple are at the forefront of a technology that can positively impact lives across the globe.[/box]

[box type=”shadow” align=”” class=”” width=””]

Newsdesk: Which programming language will dominate the AI areas? Matlab? Python? R? What’s your view on this?

Anmol: AI is no longer a commodity. With giants like Google open-sourcing their AI frameworks like Tensorflow, the future of the technology is open source. AI is no longer just a mode of programming, it is the backbone of the most successful applications right now. This trend is likely to continue into the foreseeable future.

Programming libraries built in Python and R are readily available and developer communities have been growing from strength to strength. Python has been the most popular choice for AI development for quite some time now and will likely remain so in near future. R is a very handy language and has been the favourite of the research community for very long time. The reason why these languages are popular is that they are easy to learn and execute.

Having said that, I think there are other languages like Scala and Java which are tougher to learn for AI applications but are faster and highly scalable. Also, the pioneers in the field still need the control of low-level languages like C++.[/box]

[box type=”shadow” align=”” class=”” width=””]

Newsdesk: You have strong connections into the technology community. What would your advice be to young engineers looking to come into this field?

Anmol: The field of data science has exploded so quickly in the recent past that the education curriculum has been forced to play catch up. Even though some leading universities have started offering courses in Data Analytics, Machine Learning, and AI, they are still far from the industry standards needed to work on a production level.

This is the reason why the industry is projecting a huge shortage of data scientists in near future. This is also the reason why we have tried to replicate the data science approach programmatically. We have automated our platform so that people from the non-technical backgrounds can enjoy the benefits of this technology.

Given the current climate, I would highly recommend the enthusiasts of this field to go beyond their curriculum and work on real-world projects. It could be in the form of internships or data science competitions like Kaggle. But it is important for young engineers to develop skills beyond programming and statistics. It is crucial to understand the usage and applications of the technology first. Only then should they work backwards to learn what is required to implement those.

Data science is a field which needs skills from multiple fields, namely Advanced Statistics, Computer Programming as well as Business Knowledge. Someone once said that – ‘A Data Scientist is someone who knows Statistics more than a Computer Scientist, and Programming more than a Statistician. ‘[/box]

[box type=”shadow” align=”” class=”” width=””]

Newsdesk: Please say some words for our readers in ComingTechs where we are introducing new upcoming technologies in healthcare, and AI.

Anmol: ComingTechs is a brilliant platform where people can know about latest technological trends and their applications. We are living in an era of heavy disruption and the only way we can stay relevant is to keep ourselves updated about these. I am glad that ComingTechs is collating all this information at one place and working hard to make it available for their readers.

It is very easy to get used to of our current routine and become complacent about current technologies. I will highly encourage the readers to keep an eye on ComingTechs since these guys are pushing tirelessly to update users on latest technological shifts. This is important to keep learning new things, breaking technological moulds and innovate. Stay Curious![/box]

[divider style=”solid” top=”20″ bottom=”20″]

Find new and coming technologies at our site: https://comingtechs.com. Please subscribe to our news form and add RSS to your favorite news reader.