Q&A With Anaconda Experts: How Do You Become a Data Scientist?

“I was technically oriented from a young age. I went into college for electrical engineering, focused on computer hardware and software. My ambition was to become a professor and an academic. Then, the internet boom happened when I was in grad school, and my plans changed. I discovered I was good at marrying software development and advanced mathematics together. There’s a lot of flavors to that, but it just so happened that my training meshed well with machine learning and artificial intelligence.”

– Michael Grant Vice President, Services

“I did my master’s in statistics and a Ph.D. in educational psychology. Psychology is actually all about data and statistics. In grad school, all I did was collect data, work with data, find patterns within the data, and use statistical models to generate insights from data. I just fell in love with working with data, and I wanted to continue doing that after I graduated.

Sophia YangSenior Data Scientist

“I have a master’s degree in physics and a Ph.D. in astrophysics. In my second postdoc, I spent too much time writing papers and waiting. I didn’t want to be an astronomy professor. In the meantime, I had done a lot of programming to get the job done. That’s initially what Python and all the tools were to me. They were a way to get from data to results to publishable articles. And I realized I could do something with that because it was so useful.”

– Martin DurantSoftware & Data Engineer

“Advanced degrees are certainly not prerequisites for being able to do the work. But being able to get a job is a different matter. If you can show you have the skills, then a degree is not the most important thing. Having said that, the proportion of people who get degrees is very high, and the proportion of those who go on to get a master’s degree or more is much higher than it’s ever been. So, you will be up against people who have the same experience as you and have a degree. That competition makes it hard to stand out.”

– Martin Durant Software & Data Engineer

“I’ve taken courses through Coursera and a few others. Classes in school go more in-depth on fundamentals. Online courses are short and fast; you can learn something practical, like a tool or a language, really quickly. If you’re new to data science or want to learn the basics of a new skill, these online courses are worthwhile to get started. I still learn new things all the time with online training.”

– Sophia YangData Scientist

“I’m largely self-taught — just learning by doing. But also, I wasn’t afraid to go to third-party resources like Stack Overflow and other things you can find through Google. I also took two different machine learning and AI courses at Coursera and Udacity. It was valuable to hear from an expert perspective on the field, and it also confirmed I was a good fit because of my mathematics background. I definitely would recommend taking courses like Coursera or Udacity, but you will get more value out of them with a foundation in mathematics.”

– Michael Grant Vice President, Services

“Online or supplemental training can show you are putting in the effort. Hopefully, you can learn something actionable in the process — just having the certificate doesn’t necessarily mean you learned the material. If you don’t have experience on the books or have never had a job requiring these particular skills before, it can be tough to show your experience to an HR department that doesn’t have the time to talk to you in detail before making the first cut. Taking courses or doing bootcamps are ways to show you’re at least putting in the time and effort to learn relevant skills.”

– Martin Durant Software & Data Engineer

“A certification in pragmatic marketing was recommended to me when I became a part of the product management organization. The training gave me insight into how the organization and its teams need to operate to deliver a software product, which has been a valuable skill. My particular role is aligned with a product development team, rather than a data science team, so this training gave me an onboarding into the product management world, so I could better understand my peers.”

– Albert DeFusco Data Scientist, Product

“Strength in programming goes underappreciated. People who have gone through data science bootcamps don’t necessarily come out as particularly strong programmers if they didn’t already start as one. The nice thing about being a really solid programmer in the field of data science is that you are extremely adaptable.”

– Michael Grant Vice President, Services

“Producing visualizations is easy, but doing it in a way that means something is much harder. It’s hard to teach, hard to learn. You need to understand your audience, and you need to understand your data — separately from each other. You can then use those together in a magical way to produce something to tie the audience to the data. Being skilled at data visualization is a good way to get a job. How can you show yourself to the world? A good, understandable visualization captures attention.”

– Martin Durant Software & Data Engineer

“To me, there’s a hierarchy between these technical skills. For example, machine learning almost has mathematics and programming as a prerequisite. You’re not going to get as far as you should in machine learning without the foundation of math and programming.”

– Michael Grant Vice President, Services

“In my postdoc, I did a lot of programming to get the job done. That’s initially what Python and all the tools were to me. They were a way to get from data to results to publishable articles. I realized I could do something with that because it was so useful in its own merit. I had always tried to produce good code for my own sake, but I noticed that this was not that common of a trait out in the world. The combination of somebody who understood data and the statistics of science and could actually put these things into code — that’s what set me up to be in data science.”

– Martin Durant Software & Data Engineer

“I had been looking for jobs in the software engineering space for about a year with not much luck. It wasn’t easy for me to market myself to those positions from my academic background. I then took a completely left-field approach to the industry by leading training and customer education. Anaconda gave me an interesting position to teach basic programming and general scientific computing. It gave me the unique opportunity to learn more about data science on the job.”

– Albert DeFusco Data Scientist, Product

“Transitioning to a data science role was a natural fit. I had a basic understanding of data science approaches from exposure in academia and my own experimenting. It was that in combination with what the specific role required. In this case, it was building a machine learning malware-detection product. I checked off all of the other boxes regarding my experience: product engineering, cybersecurity understanding, etc. The role was 90% data engineering and 10% data science. Many of my software engineering skills were ready to rock, and it was really sprinkling a little data science application on top. I was able to learn from the data science team around me; they gave me pointers to fill in the gaps along the way.”

– Matthew Brock Principal Engineer

“I made the jump into data science when I was working as a solo consultant. When you’re a consultant, you try to take jobs right up your alley, but you find yourself taking a wider variety of work. At one point, I was hired to develop custom machine learning algorithms. Even though I wasn’t a machine learning practitioner at the time, the mathematics underneath the hood was a great fit for me. It was also my jump into Python at the same time.”

– Michael Grant Vice President, Services

“When you are applying for a job, you can actually point to something with open-source projects. It’s all public. You can say, ‘These are the real, material changes that I’ve made to these particular projects. These projects are important to the data science and machine learning world.’ That’s excellent proof to show that you know what you’re talking about.”

– Martin Durant Software & Data Engineer

“The people that will have the most success pivoting into data science from another career path are those that can afford to do it incrementally. In other words, you can learn programming or machine learning and then find ways to apply it to your existing domain or role. Be an agent of data science in your team or organization. It’s more of an evolution than a major jump. You can crawl before you walk before you run. Don’t start with the most advanced AI deep learning algorithms. Start with good data engineering practices, visualization, and reporting before tackling basic machine learning. Find ways to incrementally layer on data literacy and see how far you can go.”

– Michael Grant Vice President, Services