When Lou Zhang was laid off two years into his first data science job, it turned out to be the best thing that had ever happened to him. Using the interviewing skills he’d honed alongside his mentor at Springboard—Eric Rynerson, the second data scientist to be hired at Instacart—Lou secured not one but three job offers—one from Wayfair, the University of Michigan, and another from MachineMetrics, a pre-series A startup headquartered in rural Massachusetts. Knowing that SaaS startups were on the rise, potentially offering plenty of room for a data scientist to grow, he decided to take the riskiest option and relocate for a job he wasn’t sure he’d still have by the end of the year.
“The first couple of months into this job, I stayed in an Airbnb because I was convinced that this startup wouldn't last very long,” Lou recalled. “But lo and behold, three-and-a-half years later, we're 15 times that valuation and still here.”
Now he’s the director of data science at a not-so-small startup that builds industrial IoT (Internet of Things) platforms for manufacturing machines. Machine shops and manufacturing plants are still “mired in the 20th century,” says Lou, with machinery that has to be manually operated by a technician. Installing IoT capabilities enables these machines to be activated and deactivated remotely, not to mention surfacing data insights that make it easy for maintenance crews to fix the machines before they break down by anticipating machine failure using predictive analytics.
This movement in industrial IoT is known as “autonomous machining” where machines predict and diagnose issues and automate the necessary changes.
“It's not all about getting that last 1% out of your machine learning model,” Lou said of what it takes to be a good data scientist. “It's more about thinking outside the box where other people don't have the skillset to make sense of billions of rows of data.”
When I started at Springboard, I was working for a trade association that represents manufacturing companies in Washington, D.C. I didn't like that job so much, frankly. The manufacturing industry was interesting, but trade associations are these slow-moving bears as far as organizations go. I wanted to see what else was out there, so I got in touch with Springboard.
My company at the time offered to sponsor tuition for me and two of my colleagues, so it was great to be able to do the program with other people. I picked up a lot of skills in a very short amount of time and I got a lot of context about what it was like to be a data scientist from my mentor, Eric Rynerson [data scientist at Instacart]. So it wasn't just like a master's program or some of these other boot camps where you just do independent work by yourself.
In the end, I landed a job as a data analyst at IHS Markit even before I finished the program. I credit the approach that Springboard had, which was to get in there real quick, get projects under my belt, build a solid foundation, and a portfolio that I could show off to employers.
I worked at IHS Markit for about two years and I really enjoyed the job, the team, and everything at that time. I also came back to Springboard to become a mentor once I had some experience under my belt. One day, my company hired a new chief data scientist based in London and he decided to move the entire team over there, which meant that I no longer had a job.
I was pretty panicked because I didn't know what to expect from the job market at that point. But my mentor had helped me hone my interviewing skills and understand what people look for in a data scientist. It's not just about achieving the highest accuracy in your model; it's thinking strategically about business problems. I was able to get, not one, but three offers from Wayfair, the University of Michigan, and MachineMetrics—which is where I am now.
I decided to go with the riskiest option, because why not? So I joined MachineMetrics, which was this pre-series A, 20-person startup in the middle of nowhere in Northampton, Massachusetts. I was able to add value quickly and build products that shipped to customers within my first couple of months there using skills that I'd obtained at Springboard. I worked with millions and billions of rows of SQL databases and was able to really make a practical impact.
After about a year of being there, in 2019, we raised our Series A seed funding. Thanks to the contributions that I had made—and I feel very fortunate for this—they promoted me to lead data scientist.
Then we came up with a product with an extremely high profit margin, which was capable of predictive analytics. In the world of machining and manufacturing, that means predicting what machines are going to do before they actually do it and either stopping them before something catastrophic happens or making small offsets or things like that. That was a whole paradigm shift for the company—to not just read data off machines, but to actuate back to them, which helped us raise our Series B funding as well. So earlier this year as we closed our Series B round the company promoted me to director of data science.
Yeah, absolutely. Now, I lead a team of five data scientists and I'm honored to be able to do that. And really I credit this meteoric career trajectory to my getting started with Springboard a couple of years ago. I think that this would have taken much longer if I had gone through a formal Master’s or Ph.D. program. Springboard is all about getting a lot of experience by doing a variety of projects from NLP to signal processing to machine learning so you're not pigeonholed in any one place for too long.
IoT is a really interesting field, especially industrial IoT. This is where we’re starting to see data science and machine learning influence [physical objects] outside of the digital sphere. It's not just about showing you the right ad or giving you the right movie recommendation on Netflix anymore. It's actually creeping into our physical world.
On a daily basis, I talk to customers about their physical challenges with machining. Are they creating too many scrap parts? Are their tools breaking? Do their machines need to be stopped at a certain time, but they don't have enough personnel to program them? My day-to-day is about finding these cyber-physical challenges and solving them with the power of big data and algorithms.
I spend about a third of my time talking to customers—external customers like machine shops. The other third of the time I’m working with code, programming in R, Python, and PySpark to build tools to solve their problems. The rest is general coordination—leadership and product management duties where I coordinate internally with my team and other teams to get this thing built and deployed on our edge devices. It really requires a generalist mindset where you're not just a coder or a salesman or a product manager. You have to be all three at the same time.
That's correct. And I think it's going to become more in vogue as we move further along with edge computing [the deployment of computing and storage resources at the location where data is produced]. Factory floors have been mired in the 20th century for a long time—they run basically the same as they did in the sixties and seventies, especially small mom-and-pop machine shops. Now, all of a sudden, you can see trillions of data points, meaning you don’t have to have a person stand next to the machine and babysit it all the time.
It's really cool to actually see our work manifest in a physical sense. We can actually stop machines thousands of miles away from our living room.
Being able to differentiate yourself in any way from hundreds or thousands of other applicants is key. So find your hook. Maybe you know someone at the company, or your current company is well-regarded in your industry. If you live close to an urban center like Boston or San Francisco, go to networking events, meet as many people as possible, and set yourself up well for that first interview. When I received those three job offers, all of them came from people in my network. My mentor at my previous company knew the co-founders at MachineMetrics and helped get me an introduction. Out of the blue, the CEO called me and said, "Hey, I'm downtown, can you meet me?" So I rushed down there in my sweatpants.
During the interview, make sure that you're acting like yourself. When I'm recruiting for data science positions, I know that the technical stuff is not the hard thing to teach. Anyone can pick up Python or R within a couple of months, especially with all the tools and resources available on Stack Overflow. It's more about if you're a person I want to work with. So stay relaxed and talk like a human. If the company is interviewing dozens of other people for the job, then the technical knowledge is probably commodified at that point. So try to differentiate yourself early on.
That was a great experience. I thought it was important to come back and give my perspective both as a previous student and a current data scientist. I think it was really unique to be able to do that and to show current students that people actually do finish the program and get meaningful data science jobs.
I think it’s about being confident in your abilities. Knowing that you have the wherewithal to be an effective contributor is important. Keep in mind that you were able to complete the course at Springboard and get the job for a reason—they didn't just hire you out of nowhere. Getting rid of imposter syndrome is hard, but it's essential to comport yourself as an effective data scientist.
You can help determine when machines should stop and help people save thousands of dollars in scrap parts. Your work is actually going to make a difference in this world for the better. And I think that's also a powerful motivator to assuage some of that fear that comes with landing your first data science job.
Since I was the first data scientist at MachineMetrics, there were no established data science pipelines. All the code was written from scratch essentially. So they had big databases for me to play with—PostgreSQL, AWS, and S3 databases. But past that, the world is your oyster. You have to really get into it to figure out how to derive value from the data.
Having this big, open field to play with is really fun, but also really scary. You have to hold up well to uncertainty and keep an open mind as to what holes the data may have and maintain general skepticism about what people tell you about the data.
Corey applied to an open position at the company. I found out later that he was from Springboard. Springboard is really the gift that keeps on giving. At this point I've referred a couple of people from Springboard, and from what I can tell, they've all gotten meaningful data scientist jobs in various industries. You don't even see that from top Master's programs.
Bar none, it was one-on-one mentorship. My mentor, Eric, really influenced my decision to join Machine Metrics because he had an inside view about what industries were about to explode. That’s why I joined this tiny SaaS company in this burgeoning industry that not many people had heard of at the time.
I think one of the things that Springboard does well gives you the full technical picture. I would also advise that data science isn't about technical problems a hundred percent of the time. You're paid a six-figure salary because you're supposed to be making a business impact that is worth multiples of your salary. Think about problems in the sense of what value they're going to create such as cost savings or creating new products you can charge money for. As a data scientist, you're there to add business value.
Learn as much as you can about the business and the industry. It's not all about getting that last 1% out of your machine learning model. It's more about thinking outside the box where other people don't have the skillset to make sense of billions of rows of data.