Benjamin DeHass | Fremont, CA
A rigid constraint narrows the feasible region. This is a fact I have come to know intimately through coursework, external research, and general application of my education. Similar to early algebra or sounding out vowels, it became internalized. The problem with some internalization is the compartment it becomes restricted to as a result. I only applied it to external systems, to models, to problems with clean edges. When I stumbled on a rigid constraint planted in my own life, I had to consciously overcome the pushback to do anything about it.
The choice to begin an internship in January is not entirely rare, but it causes logistics a person must face before accepting. The work runs longer, especially when extending into two rotations as I have. School is foregone and set aside for the duration. Timelines shift. Decisions downstream are altered. In the rigid version of my timeline, I go to school August through May. Summers are for internships and odd jobs. Everything stays on track.
By electing to go to California in January, I accepted several things at once. I would no longer graduate alongside friends I started with. I would leave my family for the longest continuous stretch of my life to that point. I would need to learn a completely new way of living, in a place I had never been, with people I had not yet met. These were the costs of relaxing the constraint. They were real and I felt each one.
Had I kept it rigid, I never would have been caught in a hailstorm hiking Half Dome. I never would have climbed Mission Peak on a Friday evening with nothing but two water bottles and one headlamp between three people. I would not have met the people I met, worked on the problems I worked on, or learned what I learned about how I operate when the familiar scaffolding is gone.
I arrived in January with a plan that extended beyond the internship. For months prior, I had been researching and designing a personal scheduling tool, one I wrote about publicly in Speaking Wave. The idea was a weekly planner that would learn my energy patterns over time instead of just following whatever I told it. Schedulers obey input. They do not notice that every Tuesday at 2pm the deep work block fails, or that the week after travel is always slower than the week before. I wanted to build the one that noticed.
The project was a focus of my research for months. I read papers on memory systems, studied agent architectures, spent evening hours outside of work sketching how the thing should function. It was mine, built for me, solving my problem.
A few months into the internship, I was debugging a data pipeline for a stakeholder new to me. The former owner had left the company. I was rebuilding my understanding of the full system from raw tables through to production code, one connection at a time. This was not the first time I had done this at work. It was the first time I recognized the pattern. Every departure, every team transition, every rotation carried the same cost. The knowledge that made a system legible walked out the door with the person who held it. Whoever sat down next started from something close to zero.
I had been building a net to catch pieces downstream. The real problem was upstream. The scheduling failures, the repeated re-learning, the sessions that started cold, these were symptoms. The knowledge itself was not being retained by the system. It lived in people, and people leave.
The research I had been doing for my own project found a better adjacent use at work. I began building a knowledge graph that mapped the data pipelines, every table, every transformation, every dependency, into a structure that could be queried in seconds instead of reconstructed over days. What once required sitting with a former owner and absorbing their memory of a system could now be surfaced by asking the system itself.
It grew past me. New analysts coming in behind me would not need to start from zero the way I had. The tribal knowledge that existed only in heads was now encoded in a structure that outlasted any single person’s tenure. The pipeline I had to retrace manually months earlier could now be understood in a fraction of the time.
There is a version of this story where I held onto Speaking Wave as originally designed, kept the evening hours for my own product, and treated the work as separate. That version is rigid. It protects the original plan at the cost of what the plan was actually trying to accomplish.
I did not set out to build something for work. I set out to build something for myself and found that the problem I cared about had a larger expression right in front of me, on a team that had taken me in and spent real time helping me grow as a contributor and a person. Leaving behind a system that serves them after I am gone is not a departure from the original goal. It is the version of it that mattered most, and it only existed because the constraint was allowed to flex.
Seven months is not a long time measured by a calendar. Measured by what changed, it was one of the densest stretches of my life. The feasible region turned out to be larger than I thought. It usually is.