The Interview with the Hiring Manager series from Intulog aims to provide aspiring data science, engineering & analyst candidates a glimpse of how potential hiring managers view the field. In this series, Hiring Managers share best practices, tips & feedback for prospective candidates.
This week, we speak with Ashot Mkrtchiyan1, a 20-year veteran of the data science field. Ashot started his career in managing logistics and operations for a few manufacturing companies and gradually transitioned into operations analysis, process optimization, advanced analytics, reporting and BI while managing various analytical programs and teams at several well-known global companies. He has worked as a Senior Manager for Cisco’s Engineering and Worldwide Partners organizations and eBay’s Experimentation Analytics team and - most recently - as a Director of Strategy and Marketing Analytics at TE Connectivity.
Could you tell us a little bit about your career journey and how you got into data science?
My educational background is in engineering and the last degree I got was in Industrial Engineering / Operations Research. After school I got into industry and was involved in managing operations, logistics and other aspects directly related to production output. I was very happy with the day-to-day work but realized that my interests were more aligned with looking for ways to make the output more streamlined and productive using analytics – so I was on the constant lookout for problems that fit these needs all the time.
I did quite a bit of data visualization, production and process modeling and supply chain optimization. Eventually, it became so interesting that I got into reporting, analytics, and business intelligence and started a consulting engagement. The last 20 or so years of my career have been really devoted to data science – even though it wasn’t called data science at the time. In many ways, you could say that I grew up and evolved with the entire data science field.
Is there a key learning from your career in data science?
The biggest highlight for me was that how you look at data matters. If you look at the right data – and it doesn’t have to be terabytes, it could be a small dataset – and you understand what the dataset is telling you, you can use various analytical techniques and get interesting insights that other people may not. On the flipside, you can have all the data in the world, but if you don’t have the right analytical vision to correctly extract the real underlying story, you might not get any reasonable insights.
One needs a good analytical foundation and vision to be a good data scientist, not just know the latest techniques.
As the field has matured and your own data career has grown, what are some of the key changes you see in data science?
As we move into a more connected world, the biggest change I see is the shift toward heavier use of data science techniques and processing more data with certain algorithms to produce more analytic outputs, as opposed to applying critical analytical thinking to understand what the already existing data is telling us. That has become so popular that people sometimes substitute the proficiency with techniques for the analytical background and thinking.
On the positive side, if you know what you’re doing and use the knowledge the right way, you can get much more precise and scalable answers. We have a lot more data than ever before, algorithms and new approaches are getting vetted at a rapid pace. In addition to this, we have access to computational power to solve problems that we would never attempt to solve even 10 years ago.
What are some of the most exciting and challenging aspects of building and managing a data science team?
The most exciting and helpful part of a data science team is diversity in backgrounds. That way, folks on the team suggest interesting directions and come up with novel ideas. To achieve that, you have to create a collaborative environment where people freely contribute and everyone’s voice is equal. In my team meetings, I would always articulate that all of us were equally contributing team members but, in addition, I just happened to be responsible for leading the team, too.
The challenging aspects arise when someone on the team doesn’t like what others are doing and when there is a competitive team dynamic. But, again, that could be prevented by creating a truly collaborative and transparent environment and making sure the team members know what everyone else is working on and understand that everyone’s contribution is equally important.
In terms of building a team, one of the greater challenges I faced when I was hiring my first analytics team. Some of the folks were part of the larger legacy team before joining the newly created one. Some of those folks fit perfectly, but others were not at the right analytical level. So, it took a while to reshuffle the team and find the right folks within the organization. Then we started hiring people from outside. Some of the new hires were just perfect, but with some others, it took two or three months to realize they were not contributing the way we had envisioned. So, it took almost a year to build the team and get it to the level we really wanted. And a year is a long time in a corporate environment – a lot of things change. The powers started shifting, and it became harder for us to continue what we were doing, even though the team was perfectly staffed. But eventually we were able to prove the value of our analytical output.
What are some of the most important lessons that you have learned about attracting and retaining good data science talent?
Again, the basics is what I look for. The ability to think with numbers. It’s not so much what algorithms they know or how they can code. It’s their ability to see the big picture and their ability to crystallize what they see into a digestible summary to share with others in such a way that people understand the correct message, not try to guess it. That’s the key.
In terms of retaining talent, if you are at a good company that takes care of people, that really helps. The other part is, as we spoke, the collaborative environment. If the environment is not collaborative and there are compartments and team members do not communicate regularly, most of the times the output of the team will be subpar.
What are some of the missteps companies can make that lose them valuable data science talent?
If the leadership team doesn’t see the big picture, they are likely to lose good analytical thinkers. These days, it is popular to talk about data and analytics, whether the leadership team really understands its value or not. Companies hire data scientists when, in reality, the culture of the company doesn’t support it. As an example, a company I consulted with had a strong sales culture and a really good pulse of what was going on with their sales teams. Some of the folks were trying to use analytics to figure out new directions to move in and everybody on the leadership team was very supportive. But the moment the sales went down and it came to doing things, they did it the old way and all analytics budgets were cut. The analytics culture needs to begin at the top of the organization.
What about in the hiring process itself?
I think a good hiring process develops when you have someone on the recruiting side who is capable of understanding correctly the technical challenge. Sometimes you get an HR person or an applicant tracking system screening out candidates based on keywords in resumes. Those systems do work, but I am not sure that those algorithms always produce the right results. A lot of times, keywords do not paint the whole picture correctly. If a good recruiter brings you 20 or 30 resumes a week, you do get a good pool of candidates to look at. But, if an applicant tracking system is giving you those resumes, the quality could be considerably lower.
When you look at a resume, what are some key things you hope to find there?
The first thing I am looking for is signs of the ability to understand and solve real-world problems. The second thing is whether that ability is supported by the work experience. Again, the work experience is not just about writing a code here or there. It’s the ability to translate real-world problems into the analytics. The third is whether that person is capable of presenting the work they did, explaining what was done and communicating that solution properly. The fourth is whether they have the ability to use a variety of tools, because a lot of times, a person knows one specific tool and try to use it everywhere. Fifth would be their ability to work with a team and influence it in the right direction. And finally, I look at the educational background and whether it supports or augments the analytics experience – it is nice to have, but not mandatory.
When it comes to interviewing a prospective candidate, what are you looking for?
Describing situations where the candidate was able to make a difference – that really tells me something. If somebody is capable of not only understanding the real issue and properly formulating the problem, but also being critical about the common way the problem is viewed, that really shows that that person is mature enough to be able to think independently and use analytics to support their independent thinking. In that case, even if they don’t know all the tools, it doesn’t matter because they can spend a few weeks and learn how to use a new tool if they understand what they’re doing.
What advice would you give to a prospective candidate who wants to know how best to prepare to interview with you?
Along the same lines, I would tell them to look at their background and find where they were critical in their career. Often, those points emphasize the right qualities. For instance, I was once interviewing for a position and was asked to tell a story where I used analytics to completely change perceptions. I told them how the president of a manufacturing company I was at wanted to implement a just-in-time approach and pushed for zero inventory. Because we were building highly customized systems and each system was comprised of up to 7,000 parts, our lead time to build a system went from about 8 weeks to 15-20. I analyzed parts for different systems and saw that 80% were common – so I began maintaining inventory on common parts. I also implemented a system to order system-specific parts immediately after the order was finalized. That brought the lead time down to 6 weeks. Even though my approach was contrary to what the management wanted, I used analytics to prove it and improve things.
What vital skills do you think a rookie data scientist should focus on developing? What kind of experience should they be looking to get under their belt?
It’s very important to be able to determine issues, translate them into specific formalized mathematical problems and then to solve them – hence, it is important to master all parts of that sequence. First, you observe what is happening and which parts need to be addressed. Then, you focus on how to formalize it as a mathematical (statistical) problem. Then you choose the right tools and solve it. Finally, you determine what information do you need to communicate to the others. You need to make sure people understand not only your points, but also why they were important in formulating and solving the problem.
You have to get experience in that end-to-end chain. If someone is only focusing on one or just a few links of it, they may become very good at those, but they will not be true data scientists.
What are some ideas, myths or misconceptions that you think rookies should unlearn from their data science courses or bootcamps?
One of the misconceptions is that the ability to code is a substitute for everything. Out of the links that we spoke about earlier, knowing one or just a few links is not enough. That doesn’t mean that such people can’t work in a business environment – they can, and quite successfully. But to be true data scientists, they need to be good at the entire end-to-end process.
How can candidates who are early in their careers stand out from the crowd?
Look at problems that have already been solved and come up with alternative solutions.
That would cultivate the critical approach and improve problem-solving abilities. Those candidates will likely stand out because that way they will have developed their independent thinking and skills that support that independent thinking.
Another good avenue is doing something for free, helping people and organizations with analytics. There are a lot of free tools available these days, and you can use those to help people.
Looking ahead at the next 5-10 years, what future trends do you see in terms of data science jobs?
The next few years, we will be moving towards wider and wider use of algorithms. However, at some point, the outputs of numerous stand-alone algorithms will start contradicting each other and people will realize that there is a disconnect. At that point, there will be a need for a fresh creative thinking approach and performing data analytics at the level of algorithm output rather than just raw data – the same way it was done on raw data a few decades ago, when the analytics boom was just starting.