I was a data scientist at NASA. Here are 5 things to know before you enter the field as it evolves with AI.

• Chris Mattmann worked in data science at NASA for nearly 24 years.
• He shares the five warnings he'd give others who want to break into the field.
• Mattmann emphasizes the importance of discipline knowledge, a supportive network, and adapting to AI.

This as-told-to essay is based on a conversation with Chris Mattmann, a 44-year-old data scientist from La Canada Flintridge, California, who previously served as NASA Jet Propulsion Laboratory's chief technology and innovation officer and division manager of artificial intelligence, analytics, and innovation organization. Mattmann spent nearly 24 years at NASA before joining UCLA in June 2024 as chief data and artificial intelligence officer. The following has been edited for length and clarity.

I got started in data science long before it was even known as "data science." When I studied at the University of Southern California from 1998 to 2007, I worked on data architecture, data engineering, databases, and data systems. My biggest interest was how they were all interconnected.

I started working at NASA as an academic part-time employee in January 2001. Soon after, I was hired full time as a data engineer and software engineer.

I moved up at NASA's Jet Propulsion Laboratory (JPL) by working on missions, and had my big break while working on the Orbiting Carbon Observatory Mission, a next-generation earth science instrument. I became JPL's chief technology and innovation officer in 2020.

Here are the 5 things I'd warn anyone who wants to get into data science

1. Study the discipline or data field that you'll be working in.

When I entered the industry, I had a lot of training in software development and engineering.

I saw that people coming out with data degrees were more effective as data scientists than the software developers who took years just to learn what latitude and longitude were; teaching an earth scientist Python was more effective than teaching a software engineering Ph.D. earth science.

Looking back, I wish my first five years were spent learning earth science, planetary science, and more math rather than software development or engineering, which I could've picked up in greater detail later.

I recommend that folks get a discipline science degree rather than a computer or software degree. AI is coming for your software engineering job, but it isn't the best at discipline sciences. Getting a degree in those areas will allow you to have a lengthier career.

2. Early in your career, get some experience with data science and AI operations.

Someone can enter this field in two main ways.

The first is by doing something interesting with open-source tools and data and then putting it on your GitHub for others to review and see, which proves you can do a real-world problem. Kaggle has many challenges like this where you can compete against others.

The second is by working or studying under a mentor or doing an internship, where you make a publicly reviewable contribution to data.

For me, the sweet spot is to carefully navigate both data research and operations; don't just hide in the research domain; instead, actually learn the software engineering necessary to deliver the application of research, data, and AI to customers.

I find operations to be a much more rewarding and less cutthroat field than research and the science publishing community and pipeline. My work in operations has included everything from NASA missions focused on big data for earth science, to software like Apache Hadoop and Apache Tika, which are used around the world by tens of millions of people and companies.

3. To succeed in data science, prepare to be considered "the help" rather than the person driving the domain.

The biggest challenges were preparing myself not to be in the lead, spending time working in the background on data analysis, and having others get most of the credit for the work I enabled.

I often say I'm a "little s" scientist, not a "big S" scientist like the others — mostly because I was made to feel that way. I was constantly put as the second or last author on papers that I was the largest contributor to, so the earth or planetary scientist could get their first author paper at a major conference.

You're an "assistant" in many cases to the actual discipline scientist that you're preparing, analyzing, and working on the data for. You're also now the "help" for fueling AI, which is heavily data-dependent, so you have to humble yourself greatly.

When you have a manager you really trust, you can try to move into data, AI research, and analytics, so you can share the credit without being worried about getting pushed to the side.

4. Build a network of friends to support you through your data science and AI journey.

Community in data science is so important. You need folks around you to lift you up and be interested in what you're doing.

Data science is definitely a team sport. While I did work with some scientists who belittled my work, I overall had a supportive system of people around me, including my family and my peers. If you don't have this and feel isolated, then get involved in data science competitions, go to meet-ups, and build your friend network and community.

Before deciding to pursue a career in data science, consider whether you enjoy analytical tasks or operations work and whether you prefer being a team player or a leader. A mix of both analytical and operations skills tends to lend itself to good leadership in data science.

Be aware you can easily burn out in data science and get bored. This is especially true if you spend your time only in Juptyer Notebooks and data analysis. Having a support network helps to avoid boredom and burnout.

Making sure you have the opportunity to move between both sides of the data science pipeline — operations and research — also helps to avoid this burnout.

5. AI will change the field so much that software engineering will no longer be as important.

Data analysis will be replaced in the next five to 10 years, done better by AI than humans. But training new AI and refining data will still have a big place and you can focus on that.

AI is data-hungry for fuel, and understanding math, statistics, and how to evaluate data science and AI will be much more important than building it.

Understanding the legal and ethical implications of building AI and data models will also become very important. People are so overwhelmed with data and misinformation that you'll have to prepare to tell better data stories and to be a background player in AI.

I'm working toward overcoming these challenges by thinking about how the job market will change. I see where the winds are headed and I'm getting ahead of them.

There's still a high demand for data scientists

Data science across disciplines is certainly something that's been a calling for me.

You can find it nowadays across industry, government, commercial, and academia sectors. And there's still high demand for the skill and profession even in the age of AI because data is the fuel for AI.

Despite the recent DOGE cuts, I wouldn't mind being a data scientist at NASA now. Data science is one of the few fields resilient to the current federal budget pauses and reductions. Being a data scientist positions you well, given this new government direction.

Do you have a tech story to share? Contact this editor, Jane Zhang, at janezhang@businessinsider.com.