How Machine Learning is Impacting E-Commerce @AmeenKazerouni @Zappos (Episode 46) #DataTalk

Listen to the podcast:

Every week, we talk about important data and analytics topics with data science leaders from around the world on Facebook Live.  You can subscribe to the DataTalk podcast on iTunes, Google Play, Stitcher, SoundCloud and Spotify.

This data science video and podcast series is part of Experian’s effort to help people understand how data-powered decisions can help organizations develop innovative solutions and drive more business.

To keep up with upcoming events, join our Data Science Community on Facebook or check out the archive of recent data science videos. To suggest future data science topics or guests, please contact Mike Delgado.

In this #DataTalk, we talked with Ameen Kazerouni (Data Scientist at Zappos) about ways machine learning is changing e-commerce.

Here is a full transcript of the interview: 

Mike: Welcome to Experian’s weekly Data Talk show talking with some of the smartest people working in data science today. Today we’re talking about machine learning and how it’s impacting e-commerce with Ameen Kazerouni, who serves as a data scientist at Zappos. Did I say your last name right?

Ameen: Close enough. No worries.

Mike: Ameen has a very impressive background. He earned his Bachelor of Science in computer science with a minor in biology at the University of West Georgia. He then went on to earn his Master of Science in computer science with a concentration in computational life sciences from Emory University, he’s now on leave there, but very, very active in the data science field. Ameen, thank you so much for being part of our data talk today.
Ameen: Of course. My pleasure. Thanks for having me.

Mike: Ameen, tell us a little bit about your journey. What led you to start working or start wanting to learn about computer science? What led you into the field?

Ameen: I started my undergrad at University of West Georgia pretty young in a program called the Advanced Academy. It was an early college program. It was a cool opportunity, where we took a lot of people from all over the world and we all lived in a dorm together, kind of just studying interesting topics, inspiring each other. It was pretty awesome.

The computer science department at UWG was phenomenal in the sense that it had amazing faculty who provided some great opportunities to explore different projects in cutting-edge areas. My adviser there, Will Noyd, and chair of our department, Dr. Adel Abunawass gave me an opportunity to take electives in robotics, which was all a craze back in ’09.

Mike: That’s awesome.

Ameen: It was like, “We should be doing robotics. That is the future.” So, those electives led to learning about intelligence systems, and I remember there was a project that we worked on where we built intelligence system using CLIPS, which is language NASA developed.

Mike: Wow!

Ameen: It was old-school, but it was a lot of fun. And I think that’s when I realized that robotics aspect of stuff was a little not my … It wasn’t boring; it just wasn’t my jam. And I liked intelligence systems and the numbers part of it.

So I was always interested in biology, and then Emory had this school program. It was the math and computer science department in conjunction with the School of Medicine. And I was like, “This sounds nice.”
I didn’t even know those were a thing. So I went there and we started under Dr. Ashish Sharma at Emory. I think medical data, again, provides so many different modalities. You have imagining data, pathology data, clinical data, so you got natural language processing, image processing, so many different opportunities. So it was an organic shift from computer science into data science.

I never got a degree in machine learning or data science, per se. Those are also all the craze. Everyone’s got a master’s.

Mike: Right, they’re all the buzzwords.

Ameen: All the buzzwords, yeah. But back then it was more computer science, computational life sciences.
It was like, “How do you define this?” It was kind of organically shifted into all my classes being about data and machine learning, and I’ve been doing it ever since.

Mike: That’s awesome. When you were at school, obviously you had an interest in the sciences, biology, and like you mentioned, computational life science. So, were you thinking about, maybe, entering the medical field?
Ameen: Oh, absolutely. Full disclosure, tried and failed. Both my parents are surgeons

Mike: Oh, okay.

Ameen: It was like, that’s what I should obviously be doing. I’m very, very grateful to whatever power got me from getting into med school because I love what I’m doing now, and that would have been a horrible decision.
That was definitely interesting, but I think the hybrid between the medical field and data science, at least while I was getting my Ph.D., it was great, very eye-opening. Now, of course, I use the same knowledge to work at Zappos.
It’s amazing how project planning the field of AI is. You can take it from one domain, stick it in the other. I think that’s what’s fantastic about it. You can keep growing in it no matter what the main area you’re working in is.

But yeah, the goal is to eventually get back to the AI mission in the medical space, for sure.

Mike: Yeah, there’ve been some amazing advancements. I read an article recently about how through machine learning it’s helping to look at x-rays faster than a doctor could, right?

Ameen: Right. And it’s crazy. It seems like stuff like that is very new and it’s amazing … I think my mind was blown when I was studying it as a graduate student. We were doing very similar things with nuclei detection in what’s known as whole slide images, which are super ultra-high-resolution pictures of histopathology slides.
Yeah, this research has been going on for years.

Mike: Wow.

Ameen: And it’s amazing the progress that’s been made. As usual, I think AI machine learning is meant to augment professionals in the domains. I think the human touch is always critical to making, especially in the medical field, those critical calls.

But, think of it as a very, very intelligent assistant device that is definitely going to help propel us in the future, for sure.

Mike: So, obviously, you didn’t take specific classes in machine learning, but you just kind of learned it along the way just because of all the work you were doing. Is that right?

Ameen: It was a mix. I would say that I did take a lot of classes in machine learning at Emory.

Mike: Okay.

Ameen: Not as many as an undergrad. I think I took one class in artificial intelligence as an undergrad. In graduate school, I took machine learning, natural language processing, high-performance computing, things like that.

It’s all a variety of different areas that are important as a data scientist. But that’s what gave me the opportunity to work with all these different modalities of data. Because the real world of data analysis and data science is very different from the theoretical. Having that option to apply on that real data was a great learning experience. It definitely helps me do my job today.

Mike: That’s awesome. So how do you define machine learning? I think we should start there before we jump into e-commerce. Because especially for those of us in business, and I don’t work in data sciences so when you’re building a business case or talking with senior leaders who maybe don’t have a background in data science, and you’re trying to propose a machine learning solution. How do you explain it in a way that is helpful for my five-year-old brain?

Ameen: That’s a question I love asking. How do you explain clustering or any particular machine algorithm to a five-year-old?

Mike: Yes.

Ameen: Because if you can do that, you truly understand it, right? So I think first off it’s reached the point where there are so many buzzwords like AI, machine learning, deep learning, big data, this, that. It’s all over the place, and everyone’s using them like a substitute for one another.

The first challenge is always finding the right word that the stakeholder likes. Because they’re all pretty much interchangeable. But when it comes to business problems, specifically, in my experience right now I found that it’s always good to include your domain experts right from the beginning.

You don’t want to bring in a machine learning scientist and say, “This is what you’re doing wrong. This is what you can be doing better.” There’s this brilliant article in Harvard Business Review. It came a couple of days ago, I think, about the misuse of machine learning.

Because different modalities can be very, very misleading if used incorrectly. So it’s important to understand the domain in which you are using machine learning. And in the process of understanding the domain, in the process of including the domain experts — and in our case it would be the people who understand fashion and the shoes and buying the shoes and work in the trenches with the product — working with them to ask how we can help and what they find difficult. And involve them with a feature engineering process.

When we’re coming up with what data we’re gonna use. We don’t use all our data at every problem. Big data is not like going fishing into every piece of information that’s available. It’s a very, very painful process of wrangling up that is going to give you the answer, right?

Fair warning to everyone interested in data science. It’s not a lie; 80 percent of your job will be data janitorial services. Guaranteed. Always the case.

So, you gotta get that little piece out, and in the process of getting that piece out if you work with the domain experts while doing it there’s a sense of ownership from both sides of the problem — from the business and from the science.

And then it’s no longer a question of how you arrived at this conclusion. We know how you did. We see the data that went in, and as a domain expert I can understand how that data, those pieces of information, leads to this answer. I obviously can’t do it at scale for the entire inventory, because it takes me time. That’s when we come in. We do it at scale. We find that signal at scale.

I think once you do that and you develop a trust relationship, you’ll find that you will have the opposite problem, where we’re not convincing people to use machine learning.

Everyone wants to just use machine learning for everything. And then you’re like maybe don’t just throw it at all problems.

Let me just run a regression or a correlation before we start using it for all that work for everything like that. So, it turned into like investing.

Mike: Are there any favorite use cases of machine learning you’ve seen being done in e-commerce?

Ameen: There are two kinds of e-commerce. There’s stuff that you see in your face and upfront. You’ve got your personalization, your recommendation, and you see websites like Walmart and Amazon, they’re all known for their personalization of it that looks cool, but then there’s also under-the-hood stuff, which is pretty great, in the sense that we have search algorithms and we have dynamic pricing, etc.

That’s very useful and very critical for the customer experience. That’s the kind of thing you don’t see out in the open, but it is critical in the journey of searching on the website and any e-commerce platform. I think it’s phenomenal and you’ve got recommendation engines run trying to browse through what shows what you want to watch. A lot of media retailers online, for example, will now not only cater to you in terms of what content they’re showing you, but also how they show you that content. There’s great algorithms out there, like even the logo of a show being used is swapped out depending on who you’re trying to show it to.

I’s not a matter of, “I know this is what you want to watch”, It’s a matter of, “I know this is what you want to watch and you may not trust me so this is the right packaging to tell you that I know this is if you want to watch”. So it’s definitely interesting, and it’s a lot of different spaces in which it’s being used ubiquitously, so it’s amazing to watch.

Mike: You talked a little bit about the A/B testing that’s going on kind of behind the scenes of machine learning and how it’s learning to predict what you might like.

Ameen: A/B testing is a different beast all by itself. A/B testing is not traditionally like machine learning. It’s more like statistical analysis, but it’s very important because no matter how confident we are that an algorithm is gonna work, the final judge of everything is the costumer. And I personally love A/B testing because it’s all about the customer.

Everything we do is to benefit the customer experience. There’s no better way to get that gratification that what you’re doing is working than an A/B test, where the customer is saying, “Yes, this is making it easier for us; this is making it better for us.”

So A/B testing is a balancing act of achieving statistical significance while getting the feature out there as fast as possible. So you see for two days it works phenomenally and business stake holders will say, “Push it out. Push it live, it’s working”. But let’s say a Sunday shopper comes in that happens to be a completely different demographic and the feature just stands over like let’s not push it out. Let’s give it some time. Let’s give it some time, let’s understand the impact it’s having. I would say that this is more of a web analytics responsibility. Data science machine learning are features that can be A/B tested, but you can A/B test funds, texts, size, colors and buttons in the website. Anything can be A/B tested and it’s probably extremely successful for various companies out there.

Mike: Yeah, I especially think about Netflix and how they’re always changing out the images of the different movie covers. So sometimes the cover … I have no interest in it, but then they change the cover and all of a sudden I’m like, “Ooh, this looks interesting.”

Ameen: There’s a lot of subliminal contact you can feed to the customer through just the computer screen, which is extremely cool. We’ve seen, for example, search relevancy is a very, very important topic in e-commerce. You are trying to find a single product like a sea of products. So we found that A/B testing different algorithms that would change the order of search results like very, very, very minor changes that you will see something go from slot 6 to like slot 4 and you see impact right there.

Hard to know why, but it’s like millions of people doing something. It’s also a little arrogant to think you know what’s going on in everyone’s head.

The analogy they really like using and kind of been my motive in life where I heard it, or whether it’s something — it’s just one of those things that I just love the analogy in the sense that I’m sure everyone knows about Where’s Waldo?, and for those who don’t it’s like a messy picture with a bunch of stuff in it and this guy Waldo and you look for him in a picture.

Mike: Yeah.

Ameen: … relating e-commerce to that because what we do at Zappos is we serve the customer, and if you think of the customer as playing Where’s Waldo?, where they’re looking for a single product in the website, my goal is not to just point at Waldo because that one takes the fun out of it and two I’m just pointing at the wrong thing, which is absurd.

The goal of machine learning in e-commerce is making that space that we should be looking for Waldo smaller. Then we put in the box somewhere putting Waldo outside the box because then we’ve kind of told you you’re looking in the wrong place. It’s kind of narrowing the scope of the problem but have fun looking for what you’re looking for. It’s shopping, you’re enjoying it, have fun but maybe don’t search through like 40,000 different products. We’ll tell you the hundred or two hundred in which your choice is hidden.

Mike: I love that. That’s a brilliant example of machine learning, and I like how you framed it with the world of Waldo not exactly pointing out where Waldo is, but here’s the general area because the shopping process is fun and that’s what you’re trying to say. You don’t want to remove the fun of shopping and choosing because that’s a huge part of the motivation to actually buy something.

Ameen: Yeah, you’re going to browse and then that victorious feeling, that I personally enjoy when I’m looking for something online. I do like to search and realize, “Oh, that’s what I was looking for.” But if it’s the first thing I see, I may go look for the right thing without realizing it’s in the sort. So you have that there but without overwhelming the customer. I’m not saying we’ve solved that but that’s kind of the Holy Grail right.

Mike: I think it’s a beautiful example, and something I never thought about, because like you’re saying, you might know what the buyer might actually want to purchase but you don’t want to necessarily put it right there because maybe they’re not ready and they want to enjoy that shopping process.

Ameen: We could be wrong, and that’s a bigger field, the bigger the bounding box the more accurate you probably are and if you come to the website and all the website was was one product just flashing like, “I know you”..

Mike: That is creepy.

Ameen: Then that’s it. We’ve already lost. There’s nothing you can do to come back from giving an incorrect recommendation.

Mike: Yeah.

Ameen: Do what you do, and do what you do well. It’s kind of a philosophy I try and stick by in machine learning.

Mike: What are some of the challenges that people working in machine learning and e-commerce have? I’m just thinking about someone who was on mobile device, then they’re on their desktop and trying to tie that person together. I’m curious about challenges you face, challenges data scientists face in this field.

Ameen: So, it’s definitely a lot more easy data, because at the end of the day customer behavior is a human behavior. I think that’s the hardest equation. Trying to understand what a person is thinking when they do the things that they do. It’s very difficult and trying to model that you might get some signal, which allows us to cater an experience that’s going to change what they do is even harder.

That coupled with the fact that it’s large volumes of data and the data cleanup is an absolute nightmare — like trying to take the data and get it to a point where it will actually work in an algorithm. There is no version like this in machine learning, that function doesn’t exist.

There’s a million steps you gotta go through to get there. Then there’s the cell, the pitch right like going to a business stakeholder they like this is the answer to your problem but also being able to say that this problem doesn’t require machine learning. So working on the right problems is what’s challenging and then just being wrong sometimes like you get an algorithm you test it out historically it just seems to be working you put it on the website and it just doesn’t do what you thought it would do. You have to accept that and kind of move on to the next thing. It is depressing but I’ve gotten used to it.

Something about machine learning is that one when you win, you win big, but you lose often. It’s a tough feeling.

Mike: I like how you talked about the failure aspect. You spend all this time working on an algorithm, it seems to be working fine, all your testing, and then you launch it live and it’s just not performing the way you thought. When you’re bringing on your data scientists, are you prepping them for the disappointments?

Ameen: So luckily with data science, it’s not … I think the disappointments come quick and early.

It’s not a disappointment as much as that it’s what makes good data scientists like data science, and that’s the challenge.

It’s math and computer science together. It’s a difficult field and it’s exciting, but it’s hard and it takes a lot of commitment to learn and stay up to date. It’s constantly changing, and it takes a lot of effort. You need motivated people in this field, and motivated people like the challenge. So the failures aren’t because of anything anyone’s doing wrong. It’s just a matter of you don’t know if the data set is going to have the answer you’re looking for. There’s a lot of work that you need to do before you get to the point where you know whether it has the answer you’re looking for or not. I don’t rephrase that, maybe not comment of failure, because now you know that that answer is not in that data set, and take that knowledge and you know what you saw on the data, and use that knowledge to like tackle the next problem.

More than anything, with data scientists I personally look for an intellectual curiosity because data science is such a little field. There’s so many different aspects, so you want to find someone who’s curious and willing to learn and willing to take a few hits and kind of just keep going with it and keep going with the problem.

Mike: I’m curious about what excites you about the future of machine learning in regards to e-commerce.

Ameen: I think machine learning is most likely gonna redefine a lot of the ways traditional commerce works. We already see the signs of it in the sense that different companies are taking on different approaches to how they cater to an e-commerce customer in sending out — you know, bosses with products preselected, etc.

And I’m excited to see AI and machine learning being democratized, becoming more easily available to as many people out there. I’m very curious to see what the power of AI and machine learning unleashed would look like when it was an Excel function, for example. And rather than needing a team of six data scientists in every company, I think my fear is that you’re not gonna need data scientists in every company. You’re gonna need data scientists and machine learning specialists just in the machinery companies, and they’re going to be producing software and tools where data science just becomes like a calculator. Only mathematicians don’t use calculators.

It’s gonna get to a point where domain experts are just using machine learning as a product of day to day work. When that happens, it’s like upgrading the engine in our innovation core. We’re gonna be propelled forward, and I’ll probably be out of a job, but that’s fine. I think it’s gonna be very interesting, very cool, to see it happen, but there’s a lot to overcome. There are a lot of fears and unknowns associated, so I’m curious to see how it pans out for sure.

Mike: We’re coming toward the end of the show where we go over the same five questions with every guest. The first one is: What is your favorite programming language and why?

Ameen: This is a fun one. The Zappos data science team is a language-agnostic. We will have you no matter what language.

The goal is for you to get the job done in a memory efficient and computationally efficient fashion without your computer exploding and it works. I learned how to develop in Java so I’m comfortable in Java, which is not the right answer. I would imagine a lot of people expect Python to be the language of choice. I’ve also run into, and again this is an opinion, which probably people get very upset like I have someone on my team who prefers Python and she and I will have arguments regularly.

But you know I’ve run into issues with large amounts of data with using Python because they have this thing called the Gil and I can’t get too technical with it. I prefer Java personally, but again I think it’s a language-agnostic situation.

Mike: When you were doing your graduate school work, did you use Java then or did you use something else?
Ameen: Graduate work was unfortunately in C/C++ because of the image processing stuff. Now, that is a language I’ve worked in and did not like it.

It was unfortunate that that was the predominant language there. Java was undergraduate and it was very painful for a while. Now it’s just a mixture of whatevers needed for any problem that comes our way.

Mike: Second question. What advice do you have for somebody who wants to enter the data science field but doesn’t know how to get started?

Ameen: There are a lot of amazing resources online, and I think one very important thing is understanding which part of data science you wanna get in and doing it right. So shameless plus here, we’re going to be opening our internship applications soon for data science. We’ve got three positions open, which I think is a good way to understand, at least that’s the way I see it. They are a data analytics position, a data science software engineering position and a data science machine learning position.

And a lot of people are gonna want to be able to do all of it, like be great at scale software engineers and brilliant machine learning scientists and amazing data analysts and statisticians at the same time. At the end of the day, math, software engineering and business acumen all bundled up into one person. That person’s a unicorn; you’re not gonna find that person all the time, and it’s probably easier to become that person overnight.

So pick one of those areas and excel at it and just understand the run. As long as you do that, you’re gonna find yourself in a team where somebody else excelled at the other thing. Take that opportunity just to learn from each other. There’s nothing more valuable than actual real-world experience, like going online and working on data sets and solving problems that are posted online. Then of course education. Go out there and just get that master’s in computer science or something like that to help you. If you can it’s valuable without a doubt and if not then it’s absolutely not necessary. I’ve worked with data scientists who are chemists, who are biologists, who are physicists from all over the place.

Mike: Awesome. We’ve got a question here. Someone’s asking if you can apply data sciences to other fields, like biology.

Ameen: Of course you can in the sense that data science is a very plug-and-chug kind of tool. At the end of the day, everything comes down to numbers. All data science is giving and taking prior knowledge and making an educated guess. One area of study that’s very interesting to me is environmental biology. They are trying to understand different impacts of certain things on an ecosystem and how they’re gonna affect the wildlife, etc.

Any of those areas as long as prior knowledge collected can be turned into a machine learning problem. I think what’s critical is understanding that it can be applied in all these fields, and it should never be attempted to replace a field with it. That’s an opinion I think a lot of people believe — that AI is gonna automate a lot of things. But we have seen attempts just automating our jobs or certain day-to-day tasks fail in the past. It’s a matter of understanding that I look at AI as an augmentation and not as a replacement. It’s meant to make jobs easier and make innovation faster rather than be a contest. It takes the best minds and puts them on discovering the new thing because AI can handle the stuff that’s already been completely fleshed out.

Mike: Great answer. So now the question is more designed for those that are data scientists or business leaders looking to hire a data science team. What would be your advice on the type of team to form and what to look for in a candidate?

Ameen: So the data science team in Zappos — I was lucky enough to be able to help build major portions of it out, which has been a great opportunity and a good learning experience for me. I think that catering to that diversity of the different skills in data science that are necessary for a successful team is important.

Understanding exactly what you want from your data science is a given. Is it data science producing a customer-facing, at-scale API for personalization? You’re gonna need some software engineers, machine learning scientists and analysts on it. Busy data science team meant to be doing ad-hoc analysis where they can use it like elite or risk management or forecasting sales. Then you’re gonna need a good statistician machine learning scientist combo, not as much software engineering expertise. So understanding what you want from your data science team, and then being able to break it down across those three skill sets, is important.

I would also recommend against this mistake I’ve made in the past, going after that recruiting hunt to find that person that can do all of it because that’s been a waste recruiting bandwidth recruitment dollars …

Then you’re wasting bandwidth and not having an extra pair of hands on the group that could have done a lot in your quest to find the unicorn. The critical thing you should look for is the ability to learn new things, motivation to actually do it when necessary. It’s weird because that’s so hard to judge them in the 15 to 20 minutes or hour you get with the candidates nowadays. But if you can find that, do whatever you can to attain it because it’s very rare and it is definitely very important in this field in particular.

Mike: What would be a question you would ask that candidate during that 20-minute interview to gauge their curiosity and drive?

Ameen: When we recruit over here we do a technical phone screen and then an onsite interview. If the person is impressive in the technical phone screen but has clear areas of gaps in the knowledge and the toolkit, what I like doing is telling them, “All right, I’m gonna ask you about those when you’re onsite.” The onsite interview is usually a week later. Then I see how in depth they were able to understand that completely new topic in that short period of time.

Another fun question I like is exactly what you said about “explain it to me like I’m a five-year-old.” It’s phenomenal because it shows a deep understanding, but you can sometimes see an excitement in the person when they’re explaining it. Like data science is magical no matter how much you understand the theory behind it — just the joy of watching the algorithm predict something I’m still like, this is so cool.

Even though I know the math behind it, looking for that excitement is critical because then that excitement’s going to carry forward to a lot of new things.

Mike: We have to get going, but I have one last question from an audience member who’s asking about how important it is in the industry to know the statistics that go behind different machine learning models.

Ameen: That’s a great question. I honestly don’t think there’s an exactly right answer that covers all jobs in machine learning. Because if you’re working in an AI company, yes it’s critical because you’re working on improving algorithms and making the algorithms we have to understand statistics. But at the same time, if you’re in e-commerce, there is definitely value in someone who has experience applying existing models to different data sets who may not understand the theoretical foundation behind every algorithm.

So it depends on the job you’re trying to do. If you’re planning on being a machine learning research scientist, there are so many different names for the same job with different varying levels of skill sets at different areas of expertise. But if you want to be a research scientist — yes, if you want to be a machine learning person who focuses on business intelligence and business analysis, it’s more important to have the proper toolkit than the deep understanding of the statistics behind the models.

It just depends on the job you’re going for. Ask yourself the honest question. Is it the right description and can I do this? If you are confident you can then it is probably not the right machine learning job for you, or data science job for you. But at the same time that doesn’t mean that isn’t one that caters to the current existing tool belt or that isn’t a couple of tools you can add to your tool belt.

Mike: Wonderful. Ameen, thank you so much for your time today. You’ve been an awesome guest. I learned a ton. I know everyone here has so many other questions to ask you. Where can people reach you to learn more about you?

Ameen: So I’m not aggressively on social media. I do tweet on occasion. My Twitter handle’s Ameen … I forgot what it was. It’s Ameen Kazerouni, and you can shoot me questions there. I try and share interesting articles that I come across in my stumbling use of the internet, but also hit me up on LinkedIn if there’s anything you wanna jump into in more detail. You can find me on LinkedIn and time permitting I am happy to get into email conversation you can jump on a call and chat about whatever you’re interested in.

Mike: Okay, wonderful. For everyone who’s tuned in or listening to the podcast, the short URL to get to Ameen’s LinkedIn profile is ex.pn/ameen. Also, if you’re watching the video either on Facebook or YouTube or having the About section or in the comments links to Ameen’s Twitter accounts and LinkedIn profile, you can follow him there. You definitely want to keep Ameen on your radar for all the work that he’s been doing.

Ameen, thank you so much for being part of our show today, and I’m looking forward to talking again soon.

Ameen: My pleasure. Thanks for having me. Anytime I can do this again, I’m happy to do it.

Mike: Awesome. Take care.

Ameen: Thank you. You too. Bye.

About Ameen Kazerouni

Ameen Kazerouni is a Data Scientist at Zappos Family of Companies and Founding Partner at Bumblebee Analytics. He earned his Bachelor of Science degree in Computer Science with a minor in Biology at the University of West Georgia. He then earned his Master of Science degree in Computer Science with a concentration in Computational Life Sciences from Emory University. He is currently on leave from earning his Ph.D. in Computer Science with a concentration in Biomedical Informatics from Emory University.

Follow Ameen on LinkedIn, Twitter, and Quora

Check out our upcoming live video big data discussions.