Speaker
Description
Research software engineers entering academia from industry bring engineering practices shaped by production environments. How well these practices transfer to academic research — where constraints are not only technical but also regulatory and organizational — remains underexplored. In this lightning talk, I draw on my experience as an early-career RSE who transitioned from software development to academia, and present a case from my first project to illustrate how industry-learned practices translate in research settings.
The project involves optimizing a health foundation model developed in a heavily regulated sandbox environment, where gaining access can take months. Rather than waiting, I engaged early with the research team to find a workable path forward: generating synthetic health data, developing and testing on an external HPC cluster, and applying the results back to the sandbox. Because sandbox resources are scarce and shared, this approach kept development moving without competing for limited infrastructure.
The strategy was not arrived at on a whim. It was informed by a project I previously led in industry, where a legacy production system was refactored by incrementally replacing its components with modular, testable code, alongside ongoing feature development rather than in place of it. Existing functionality was preserved at each stage, and stakeholders validated changes before the next phase began, maintaining a balance that addressed technical and stakeholder demands at once.
In both cases, the challenge was not purely technical. It required negotiating with stakeholders: understanding what is feasible, communicating trade-offs, and finding workable paths under constraint. These experiences suggest that the value industry-turned-RSEs bring to research extends beyond technical proficiency. Stakeholder communication, pragmatic scoping, and working within imperfect conditions are transferable skills that help bridge research work and sustainable software practice.