Interview with Jennifer Prendki, Principal Data Scientist, @WalmartLabs - Speaker at Global Data Science Conf - March 2017 Posted on : Feb 17 - 2017

We feature speakers at Global Data Science Conference March 27 - 29, 2017 to catch up and find out what he or she is working on now and what's coming next. This week we're talking to Jennifer Prendki, Principal Data Scientist, @WalmartLabs.

Interview with Jennifer Prendki

1. Tell us about yourself and your background.
My background is actually in Experimental Particle Physics, though I hesitated with a career in Theoretical Physics or in Mathematics. I chose Experimental Physics because I believe in using math with a specific goal in mind. However, I always kept a strong interest in more abstract sciences, and have at heart to bring theory and experimentation together. Hard math is beautiful in the absolute, but becomes sublime when applied to a tangible aspect of life, and gets used to explain reality.

2.  What have you been working on recently?
I did my PhD on a topic called CP-violation, which consists in observing the asymmetry between matter and antimatter. I was trying to understand what happened early on in the history of the Universe, and worked on explaining the infinitely large by measuring the infinitely small. Today, @WalmartLabs is giving me the opportunity to use that mindset and bring in the notion of precision measurements in eCommerce. The company has very large amounts of data that allow data scientists to extract subtle signals to understand the customer’s needs and expectations. These signals are eventually used to reach our many customers and make their visit to our properties, such as Walmart.com, remarkable. We have an exciting challenge in that our customers are both online and in-stores, thus requiring a seamless experience to the user.

 3. Tell me about the right tool you used recently to solve customer problem?
There is no right or wrong tool. I am a strong proponent of diversity when it comes to choosing a specific approach and technology. My team is using all kinds of different tools, from Python to Tensorflow. The same goes for the algorithms that are chosen: it all depends on the problem we try to solve, the resources, the level of uncertainty the use case can support. Sometimes, speed of execution justifies the usage of a more simple algorithm, and my experience is that it is often not necessary to use the most advanced techniques. There is no need to use a sledgehammer to crack a nut. 

4. Where are we now today in terms of the state of Data Science, and where do you think we’ll go over the next five years
Big Data has been undoubtedly one of the most common themes in the tech-specialized press over the past few years. The development of new technologies and the apparition of Quantum Computing now allows us to collect more data than ever. In fact, the accumulated data generated by humankind is currently doubling every 1-2 years, which is the very reason that makes Data Science such an exiting area. However, volume does not mean quality, and with more volume come new challenges related, for example, to noise removal. The industry as a whole is moving towards the recognition that data scientists will have to develop increasingly more accurate and scalable algorithms, and create original solutions to make data analysis noise-resistant. I strongly believe that the future of Data Science will come from this realization, which will happen progressively over the next few years. While by 2020 the worldwide amount of data available is projected to grow all the way to 44 trillion gigabytes, no amount of data will ever be able to replace the human ability to design the proper experiments, determine the right baselines for a model, or invent the next generation of algorithms. A fun fact is that in 2007 , the overall computing power of all general purpose computers worldwide is similar to the maximum number of nerve impulses executed by one single human brain per second. Human data scientists are not going anywhere anytime soon. 

5. You’ve already hired Y number of  people approximately. What would be your pitch to folks out there to join your Organization? Why does your organization matter in the world?
According to me, people have to join a specific industry because they believe this is the future, and because the problem they are trying to solve is the right one. Therefore I usually emphasize the problems we are trying to solve involve many different data sets. Contrarily to the urban legend, Retail is not only about Marketing. We actually work with Computer Vision, Natural Language Processing, Information Retrieval, Advertising, etc. The possibilities are endless. Management is also very open to engineers and scientists using any approach they think is appropriate, and offer a very good environment for creative minds, as long as their initiatives help serve the customer better. But the ultimate reason why I joined is linked to Sam Walton’s mission, which Walmart personifies to this day: Save Money, Live Better. I couldn’t be more proud of the work we, as a company and as a team, are trying to accomplish.

6. What are some of the best takeaways that the attendees can have from your "Precision Measurements In ECommerce" talk?
I hope to raise awareness regarding the complementarity of Mathematics and Big Data. I feel that many organizations don’t realize that no matter the amount of data they collect, they need scientists to make sense of it, and that they also need to make their algorithms more powerful and accurate. Big Data alone is not the answer. While more data means a better accuracy, it also implies that we are entering the area where precision can not be overlooked anymore.

7. What are the top 5 Data Science Use cases in enterprises?
I was fortunate to work in different companies and different industries, so I think that I am actually in a good position to give an answer. As a data scientist, I am looking for four things when evaluating the scientific potential of an industry: the size of the data available, the variety of the data, the diversity of the solutions that can be applied, and the mission. I think that Internet of Things and Wearable Devices are unmistakably areas of the future for data scientists because of the size of the data that they involve and the innovative solutions that will have to be created. Regarding the mission and the variety of the data, Healthcare and Defense & Intelligence rank very high. However, the one area that I see as bringing all these criteria together is Retail. Walmart has lots of surprises in store and I am, more than ever, excited and optimistic to respect to the future of my industry.

8. Which company do you think is winning the global Data science race?
Simply put, I don’t believe that there is such a thing as a Data Science race. The use of the word ‘race’ here reads like companies are competing. This is true only if you reduce Data Science to Big Data. If you actually consider Data Science for what it is, namely, the union of Big Data and Big Math, then the ability to compare companies requires the reference to a common success metric, which is not realistic. However, I will say that I believe companies which understand and promote the importance of the scientific method are a step ahead of others. And of course, that includes Walmart.

9. Any closing remarks.
Thank you for giving me the opportunity to present some of the innovative work that is currently happening in eCommerce. I definitely hope that I will make both engineers and scientists excited with the kind of work we do, and demonstrate that Walmart is undeniably moving into the future with its best foot forward.