By Tim Riser on August 17, 2020
| Devon Morris |
Highlights
- You can’t just put on your resume, “I know linear algebra, so hire me to do sensor fusion.” Everyone needs to find a way to get “context”.
- ACME provides the mathematical foundation to go into master’s programs in other technical fields, do extremely well, and quickly get context.
- Anyone in ACME is going to have a lot of imposter syndrome. We’ve probably all felt self-doubt. There are good ways to defuse and counter it.
Full Q&A
Tim Riser: I’d like to start off with you telling me about yourself. Could you share your background and where you’re at right now?
Devon Morris: Currently I work at Aurora Flight Sciences. They gave me the title of Autonomy Engineer, which is a kind of fancy software engineer. I work on “detect and avoid” applications for aerial systems, solving the complex problem of not running into other things in the sky. I did the ACME core from 2015 to 2017 and then decided to pursue a graduate degree at BYU in Electrical Engineering. After going back and forth between a master’s degree and a Ph.D., I ultimately decided on a master’s degree, which I think turned out best—especially given what’s happened in the 12 months after I finished. During my graduate program I worked in what’s called the Magic Lab at BYU. My faculty advisor was Cami Peterson. I believe she started at BYU in late 2016. During my last semester in ACME, I took the very first class that she taught. We had a good relationship and I ended up being her graduate student. Her research pipeline wasn’t fully set up yet, so we all ended up trying to figure it out together. Most of my graduate work focused on estimation of state and tracking for unmanned systems using tools like Kalman filters and complementary filters. That work was close to what I’m now doing now at Aurora Flight Sciences. My work focuses on the tracking part of “detecting”, and we’re fusing together multiple Kalman filters for tracking.
Tim Riser: What would you say that drew you to your current work?
Devon Morris: In graduate school, I really enjoyed the linear algebra side of electrical engineering and found myself best at controls and estimation. I was offered an opportunity to work as an intern at Aurora Flight Sciences and they subsequently offered me a full-time position. I work on unmanned aerial systems for autonomy. There are a lot of tough technical problems to solve in autonomy, and I don’t know if I’m like bought into a specific platform. In other words, if somebody offered me a self-driving position, I’d probably be equally interested. Amazon has positions with warehouse bots. Mapping the ocean with underwater unmanned vehicles is a big thing that Tim MacLean at BYU used to do a lot of work with. I think my happy spot is one degree of separation from hardware. I like to see that the algorithms that I design and write get deployed on hardware, but I don’t necessarily want to be fully in charge of like all the headaches with integration and deploying.
Tim Riser: What was it that got you really excited about this job at Aurora Flight Sciences in particular?
Devon Morris: They have a really good company culture. Since the recent deal with Boeing, there’s some of this “Big Brother” slowness that’s being like inherited as we got acquired. But I gort the chance to work on a program called the Autonomy Core. It was core set of modules that would provide autonomous capabilities to a variety of platforms. The team I was working on was extremely competent, the problems were difficult, and we had good direction and a lot of autonomy ourselves to choose what we what we thought was important to work on. So those were all very important factors. Unfortunately, that program reached its end of life at like February/March this year. I’m now on a different program that is also interesting but they’re just getting their legs under them right now and are going through some growth pains.
Tim Riser: What do you think ACME students would be surprised by or might misunderstand about the work you do?
Devon Morris: ACME students know in the back of their minds that they have like all the tools to do the work I’m doing, but they might have like a disconnect as to how to actually apply these them. There’s no autonomy lab in junior or senior core, right? That just doesn’t exist. ACME talked about controls somewhat, but I think if anyone sat down for a couple of weeks and worked through a textbook on Kalman filters or robotics will say, “Oh yeah, I know all this stuff.” ACME students just need the context. I think they would be really surprised at how quickly and how easily they would pick it up. In my graduate degree, many students were struggling with math. Compared to everybody else, I was not spending a lot of time on my homework, and just working on research the rest of the time.
Tim Riser: You said earlier that this internship came in out of the blue a little bit. How did that happen?
Devon Morris: The internship came through the Magic Lab. Jeffrey Sanders is an engineer at Aurora Flight Sciences and he was one of Randy Beard’s first graduate students. When he came in November of 2018 and presented to the Magic Lab, I went and printed out a resume and handed it to him. They happened to be looking for quantitative interns and engineers at that moment, so the timing was good for the connection through the Magic Lab.
Tim Riser: And what was the process by which they brought you on full-time? How does that work?
Devon Morris: Yeah, within a month of starting as an intern there, they started making those jokes like “Hey, you know, it’d be a shame if you went back to school,” or, “Hey, we really need you.” That kind of thing. They told me I was one of the quickest people to get onboarded and commit code to like the repository. It seemed like they were impressed with me. They were great about working through the logistics my wife and I had to figure out, including how I could properly wrap up my graduate degree at BYU, how my wife could wrap up her projects at Qualtrics where she was working at the time. Towards August of last year, they offered me the job and I started at the end of September.
Tim Riser: How would you sum up the theory of how Devon Morris broken into industry?
Devon Morris: I graduated from ACME, right? I knew that I had skills. And I didn’t realize the value of all of them. With regards to Python programming, git is a huge skill. People don’t realize that this annoying thing that they have to jump through hoops to do in ACME gives you enormous advantage. If you can perfect git, you’re halfway to becoming a master at your job. So, while I knew I had these skills, it wasn’t clear to me what they were because I didn’t have the context to apply these skills. While my graduate degree ended up being fairly easy for me, it really gave me that lens which I needed to see these problems and see how the math that I had learned the last two years applied. Then, the connection through the Magic Lab is how I broke into industry, I guess. Again, I don’t think it’s necessarily something hard to pick up right, but it’s a perspective that you don’t get during ACME. If you are going into a pure data science, statistician, or economics-type position, it makes a lot of sense to hop right into industry. But for my field, I felt like I needed a little bit more context.
Tim Riser: Is that the advice you would give ACME students who want to enter your field: “get context”?
Devon Morris: You need to find a way to get context. You can’t just put on your resume, “I know linear algebra, so hire me to do sensor fusion.” That’s not going to work. But as an ACME grad, you have all the skills to figure out sensor fusion. You just have to do it once or twice. For me, getting context was attending graduate school. I don’t think that’s correct for everybody—you could get context while you’re an undergrad in ACME. You can go to the Magic Lab meetings at BYU. They really like applied math students because they don’t have to teach them all this tough math. Since you already know the math, they just have to how to upload your code to the hardware, and that‘s the easy part. So, you can get context by participating in labs like that, or by doing your own side project. There are a bunch of open source data sets on computer vision, like self-driving cars and camera feeds you can use to create control algorithms or estimation algorithms. It’s something you just have to do a couple times and then you realize that “Oh, this is how all the stuff that I’ve learned last two years applies to this very specific field.” If you went into an interview for an autonomy engineer position and they said, “How would you correlate camera data? Or how would you align LIDAR data?” and you responded, “I don’t know, it’s not something that’s taught,” it likely wouldn’t go well for you. You need to acquire that field-specific lens.
Tim Riser: In addition to git, are there other skills, technical or otherwise, that you have found to be useful in your career?
Devon Morris: Git is by far the biggest one. For autonomy, you can write prototype code in Python and that’s fine. But nobody’s going to put Python code on a vehicle. You need to know like a lower level language, like C++ or C. That’s something they don’t teach in ACME, but something for which all the principles are taught, like object–oriented design principles. You’ve got the principles, but you do have to pick it up.
Tim Riser: Are there any topics you wish ACME covered?
Devon Morris: This is my perspective, based on what I ended up going into. I think we are a little light on the signal processing. For example, there is discussion on the Fourier transform, but they don’t talk about the z-transform. It may have changed a lot since when I took it because you know, they have they haven’t completed the textbooks yet. But the stuff we learn about the Kalman filter was pretty sparse and I end up mostly relearning it when I learned it in grad school. I recalled a few things but like I said, the context matters, and how they talked about it in the electrical engineering department was very different from how we talked about it ACME. Another thing: I don’t know where you’re supposed to learn it, but system design is incredibly tough. This is perhaps more of a computer science topic but designing like large-scale systems and deciding where to place responsibilities and interfaces is not an easy thing. I don’t know how you would teach that. I think it comes from a lot of trying stuff out and failing. I would say I’m not very good at it yet, but I would be very interested in ways that I could become a lot better. The engineer approach to things is to draw block diagrams, connect these blocks with lines. That’s what they consider a solved problem from a systems–engineering perspective, but all this complication comes up from implementing it. Much like a mathematician writes a theorem and says, ‘It’s done,” each process is an abstraction and each doesn’t apply 100% accurately to the world. Somebody has to take that abstraction, implement it, and that is not easy. System design is hard and worthwhile.
Tim Riser: What have you learned since ACME about like working with teams and working within an organization?
Devon Morris: What I’ve learned is that it’s 70 percent sales. It’s funny: there are a lot of times you know you have the correct idea, or at least a pretty good idea, and getting a majority share of your team to buy into that idea is not a non–trivial process. You have to do some lobbying. You have to sell your idea. You have to show the benefits of it. Sometimes if it’s small enough, you can go and do it quickly before anyone notices, and then everybody’s happy. But it will depend on how large of an undertaking it is. I wouldn’t say I’m particularly good at the sales aspect of it all, the lobbying, or convincing. I probably should have attended more of the ACME soft skills seminars, but it is what it is.
Tim Riser: You said you were working with an extremely competent team. Can you talk a little bit about what you admire about them?
Devon Morris: My comment was about my IPT Lead, the person who leads our integrated production team. He comes from Lincoln Labs, has about ten years of experience, and is just incredibly good at C++. I haven’t met anybody as good at C++ as him. It’s interesting working in Python, because he is sometimes guilty of bringing his C++ ideology to Python. Outside of that example, there’s a lot of very capable people around Boston. It’s nice to be able to work around these people who are like very smart and competent. I don’t like to be even the 50th percentile smartest in a room, because that means I have something to learn.
Tim Riser: After working for several years, do you have a more distinct picture of what sort of team or projects you like to work on?
Devon Morris: At least for the foreseeable future, I am like very comfortable being at the bottom of the org chart, quote unquote. Doing the hard work, learning the tough lessons, taking other people’s designs and running with them. There is a lot of thought that has gone into these things that I haven’t seen before. Grad school and school only prepare you so much. I think it’s not bad to spend the first few years in industry soaking up a ton of knowledge and just learning. Coming out of ACME, you think you’re pretty smart, and you are, but you speak a different language than engineers. This comes back to that context thing. You speak a completely different language, and that’s probably true of any team that has a diverse skill set and background. Sometimes you have to just sit there for a second and learn the common tongue, if you will.
Tim Riser: What advice would you give ACME students about balancing multiple priorities while in the ACME program?
Devon Morris: Balance is always going to be tough. I don’t know if that I’ve achieved any Zen balance yet. I would say that reaching completeness in projects along the way is valuable. The more things that you can do and fully complete, and know that you’ve finished it, the better. Whether it’s a personal project or a lab, having a commitment to do it right and complete it in such a way that you can show an employer or a potential employer. If you’re the kind of person who wants to make their own side projects, that’s great. Do it, get it done, and make sure it’s at a point that it’s done and at least moderately impressive. I have tons of labs in ACME, or even my final project in ACME, where it was pretty good, but I don’t know if I would show it to an employer. I think if you just took that like 10% or 15% extra time and made it like impressive and polished, it would be extremely compelling. I don’t know if that’s balance, but it helps.
Tim Riser: Do you have any other thoughts on navigating school life while in ACME?
Devon Morris: Any competent major, and especially ACME, is going to have a lot of imposter syndrome. I think we’ve all felt self-doubt. It happened to me and I think it happens to everybody. This sounds not very compassionate, but when it happens, I think the best response is to just do something. In my view, it would be like better to spend like half an hour not worrying about it and learning a little bit more, rather than keep worrying about it so much. It’s tough when you’re in that feedback loop where you’re telling yourself you’re not good enough. You’re not going to feel good enough, and it can be tough to break out of that. Figure out how to change it up for you. I think personally the best thing you can do is the thing that’s nagging you in the back of your mind that you fear the most. I think pushing into fear is an incredible sensation and is a good motivator. There are things that I know that I should be doing, and sometimes the fear of failure is the biggest impediment to doing those things. Just pushing into that fear of failure and saying, “If I fail, I fail, but I’m gonna do my best,” is a good feeling.