[template]
Research is mostly the slow work between the moments that make it into a paper. Here is what that looked like for me at the University of Toronto.
The lab
I worked as an undergraduate researcher in a systems group at the University of Toronto, supervised by one of the faculty. The group focused on building software that had to run reliably at scale.
The problem
The project asked whether we could make a certain class of system easier to reason about without slowing it down. That mattered because the usual answer traded one for the other.
My specific focus was the measurement side: how to tell whether a change actually helped.
What I worked on
Day to day, I:
- Built the harness that ran the experiments.
- Ran those experiments against a few different setups and recorded the results.
- Read the surrounding literature so we were not repeating known work.
| Phase | Activity | Tools / methods |
|---|---|---|
| Exploration | literature review and baselines | reading, notes |
| Build | the experiment harness | scripting, version control |
| Evaluate | timed runs | the group’s cluster |
What came out of it
- An internal report that fed into the group’s larger work.
- A clearer picture of where the tradeoff actually bit, including one case where it did not.
- A working sense of how research methodology differs from shipping product.
Reflections
The hardest part was the ambiguity. There was rarely a clear right answer, just a next thing to measure. What I would tell my earlier self is to write down the question precisely before running anything.