July 25, 2026
I’ve had the privilege of having worked and volunteered in many different places; given that I’m presenting a symposium on the Discourse on the Method in a few weeks, I thought I’d record some of these as frames for my beliefs.
The first work I was paid for was at my family business. In the early nineties, my grandfather had commissioned a custom DOS program to do the books at the end of the day. However, around the time Windows 7 came out, the computer it was running on began to show its age. At the time, there was not exactly a surplus of DOS programmers available to maintain legacy software. However, my dad did have a twelve-year-old running Debian on her Raspberry Pi. If I remember correctly, I ran it first through DOSBox, which let it run on modern computers again, though it lacked plug-and-play printer access. However, you could export the report into a copyable text file which could be shared in a shared folder between the emulator & its parent file system. From there, we used Win7 default printing. After two years of this, my dad realized that we could use a modern inventory & books program.
Lesson: People dislike change even if the tools they’ve chosen to use are hampering them. Take a step back to see if you’re using a flathead on a Phillips screw.
I volunteered at the Perot Museum in Dallas during high school, starting as a docent then moving to exhibits maintenance & repair. This was an incredibly fun job; my boss, Doug, was an artist & I worked with one of my high school friends, so there were always people to talk to.
A museum has a large number of visitors per day, many of whom are extremely dirty young kids who are adept at discovering unique ways to break toys. One thousand teenagers eventually overcome red Loctite! Personally, I enjoyed robot repair & soldering because while playing Grooveshark, the hours would pass much quicker. The one break in the day was teatime, where my boss and I would sneak upstairs to the offices and raid their hot chocolate supplies when they weren’t looking.
My boss finally opened the art studio he was working on; the rest of the team had rapid turnover until my friend and I were the most senior in the department despite being volunteers. These combined expectations & lack of trust made it worse to work there; I quickly found a different place to volunteer myself.
Lesson: Change will happen; you can always leave a suboptimal outcome.
Over the summer of my sophomore year, my robotics team tore the innards of a 90s RV out & converted it to a mobile robotics classroom, then proceeded to teach on it for the rest of my time in high school. This was incredibly fun work - I love teaching - and brought us across a cross-section of the United States which we would have otherwise been blind to. A good number of kids onboard had never used a computer before; some were ESL, deaf, etc. The universalization of instruction is a common debate in pedagogical circles. Though personalized education is ideal, you only have what your environment enables you to give. However, this being a team of kids who often felt stymied by public education, we didn’t want to lessen the impact of our lessons. To do so, we asked the kids what designs & objects appeared in their daily lives, practical problems which computers and 3D printing could solve, and so on. Much of what we experience shapes how we engage in the world; we ought to teach people the ways in which the tools we give help them.
Lesson: We operate under constraint. The efficient allocation and use of limited resources makes all the difference.
Theoretically this sounds interesting; however, the majority of my work here was spent digitizing thirty-year-old records before the end of the summer. I was given a room filled to the ceiling with paper and instructions to create an archival and digital system. I did a ~Fermi estimate to see how long this task would take me if I did a certain amount of case digitizations per day. This ended up taking the rest of my summer.
Lesson: Time is often the scarcest resource.
I knocked doors & made calls all summer.
Lesson: Not engineering-based, but you have an obligation to learn how to talk to others!
To get better knowledge of the chemical engineering field, I worked as a co-op in rural Iowa for nine months. Despite the conditions, I found the work very interesting & fulfilling. Though, a fifty-year-old factory is a novel environment; one of the buildings would regularly register 150°F during the summer; we had to account for various chemical dangers; factory culture was unfamiliar. A particular incident summarizes my time there -
Due to a regulatory change, my unit was investigating the composition of byproducts in various waste streams sent to the wastewater processing unit at our plant. Since the eventual output of the treatment was sent to the Mississippi, both state & federal considerations were made such that we had to make multiple samples per week. Most could be fetched at any time as our process was mostly continuous. However, a component of the impurity removal process used an ion exchanger, which is a batch-step process. The complete process took ten hours, which made it difficult to collect samples while on shift, and sampling had to be done with urgency to catch waste before it cleared for the next cycle with higher-concentration lye. Given that lye enjoys turning the fat under your skin into soap upon contact, I would watch the clock and notify a more qualified operator upon the sampling time, accompanying them to collect the sample. Most would not wear the recommended base-resistant PPE, opting for their work clothes. As I was not supposed to take samples, I wore no PPE and stood behind them.
One morning I grabbed a different operator for assistance who had been working for much longer. Unlike most, he opted to wear the recommended wetsuit for hazardous samples. When we attempted to sample the waste tap, nothing came out; we had to collect a gallon of fluid. He diagnosed the tap as having an air trap and unscrewed the strainer to let loose the air bubble blocking the pipe. Then he proceeded to open the tap. As the strainer had never been reattached, this sudden depressurization caused the hot, fat-melting lye to spray at high velocity out of the top of the tap instead of the drain point, coating him in lye and splashing on my arm. He quickly closed the tap, ran for the showers, and checked on me. In a daze, I walked back to the control room and informed my boss. For this, I was given permission to go home. Since he was wearing a wetsuit, he was completely fine & I made off with a few blisters.
Lesson: All processes are non-deterministic and should be treated as such.
An aside: I chose not to use the very public safety shower in this scenario as I did not want to publicly expose my early-puberty body to the world. When I returned the next year, I informed the company & was moved to the furthest unit from my original such that I only interacted with new people. I was also forced to use the women’s locker rooms. As this was during the Bud Light controversy, this was a bit stressful; luckily, my coworkers had no idea what trans people actually looked like. Thus, I was peppered with questions (as one of the three women in the unit) as to what I thought of these false women. Now, when I would have sworn rural areas are lovely.
My academic research was in amyloid-beta proteins starting in 2020, the loose idea being the automation of feature categorization on cryo-EM photos of these proteins. I had some experience in early machine vision à la OpenCV, so I was asked to start on this project. Given the fuzziness of these images, traditional vision was not very effective such that I gravitated toward a new project called “you only look once”. Initial tests showed promise, but we floundered upon more rigorous analyses given the lack of training data and that I had been a freelance coder with no professional support. Eventually, we created a workable feature detector. Unfortunately, the field of machine learning figured this out at the same time.
My first job out of college was to analyze power consumption for chemical plants to minimize power costs at constant production. This can be done in Excel (at least for modulation). However, to smooth energy rates requires much more market knowledge. Texas is relatively predictable as solar & wind provide a reliable supply curve as well as most of the country with natural gas, nuclear, and wind. However, the Pacific Northwest was difficult to predict due to its reliance on hydroelectric power.
While traditional engineering provides few tools to predict energy markets, I remembered my roommate’s econometrics homework that there exist seasonal regressions which may predict these prices. Knowing nothing about them, I requested time to read through an econometric textbook & make a SARIMA model. Given that a majority of energy comes from hydroelectric, I figured the input variables ought to be weather-related, downloading decades of data from NOAA & aligning it with Mid-C prices & seasonality to run this regressor. This was surprisingly successful, allowing us to hedge energy prices against rain & snowfall.
Lesson: Off-the-shelf tools are surprisingly workable; if you want to be the first to use them, ensure that they’re from different fields.