Title: Engineering Software for the Scientific Community
Abstract: Scientific software is at the heart of computational science. Well-designed software tools make research possible and make the results of that research broadly accessible. In short, good software makes impactful research: but what makes for good software, and how can we make software good?
In this talk, I argue that high-value software results from a thoughtful software engineering process. The engineering design process begins during planning, with choosing the right audience and scope of features, and continues to drive architecture and implementation choices throughout development, into the post-publication maintenance phase. I advocate for the crucial role of software engineering expertise in balancing the competing needs of correctness, performance, transparency, security, and future utility. Using examples from my work across multiple scientific domains, I illustrate how the software engineering process guides the development of software tools that become durable, extensible, impactful infrastructure for science–both today’s and tomorrow’s.