Talks
Benefits and Limitations of Jupyter-based Scientific Web Applications
Nicole Brewer, Rob Campbell, Rajesh Kalyanam, I Luk Kim, and Carol X. Song
Scientists are increasingly interested in creating standalone web-applications as computational and data analysis tools. The authors have worked with several such research groups to design, develop, and deploy such web applications that are increasingly based on Jupyter notebooks. One of the primary reasons among many to use Jupyter notebooks is the fact that research groups inheriting these applications are capable of maintaining and extending them. In this paper, we walk through the design process for one such application and discuss development environments that are best suited to Jupyter notebook development. We then explore several other applications where we employ similar design patterns. In doing so, we expound upon the benefits, limitations, and challenges of Notebook-based applications to provide a guide for other facilitators in similar situations.
Roles of Professional Research Software Engineers in the Science Gateway Landscape
Sandra Gesing
Science gateways are increasingly used by researchers and educators evident in publications and presentations at events such as eScience and PEARC. Teams around science gateway development are diverse and need a diverse set of expertise. Skills needed are around developing and/or extending a science gateway framework with backend connections to complex research infrastructure or lab instruments, enabling collaboration with authentication and authorization mechanism as well as addressing FAIR (Findable, Accessible, Interoperable, Reusable) principles. One of the most important aspects are the needs of the specific community for each science gateway on features and user interface.
Half-Precision Scalar Support in Kokkos and Kokkos Kernels: An Engineering Study and Experience Report
Evan Harvey, Reed Milewicz, Christian Trott, Luc Berger-Vergiat, and Siva Rajamanickam
To keep pace with the demand for innovation through scientific computing, modern scientific software development is increasingly reliant upon a rich and diverse ecosystem of software libraries and toolchains. Research software engineers (RSEs) responsible for that infrastructure perform highly integrative work, acting as a bridge between the hardware, the needs of researchers, and the software layers situated between them; relatively little, however, has been written about the role played by RSEs in that work and what support they need to thrive.
To that end, we present a two-part report on the development of half-precision floating point support in the Kokkos Ecosystem. Half-precision computation is a promising strategy for increasing performance in numerical computing and is particularly attractive for emerging application areas (e.g., machine learning), but developing practicable, portable, and user-friendly abstractions is a nontrivial task. In the first half of the paper, we conduct an engineering study on the technical implementation of the Kokkos half-precision scalar feature and showcase experimental results; in the second half, we offer an experience report on the challenges and lessons learned during feature development by the first author. We hope our study provides a holistic view on scientific library development and surfaces opportunities for future studies into effective strategies for RSEs.
How to Grow Diverse and Sustainable Teams through Mentorship
Caleb Jackson
The chief question I will address is: in a sea of open-source projects, how might mid- or senior-level developers best mentor early career developers? As a software engineering apprentice, I have had the unique experience of swimming through open-source projects and ultimately finding tech spaces that align with my values (open science and collaboration) and interests (biochemistry and neuroscience). During this transition time, mentorship played a crucial role in making this happen!
So often, tech spaces lack diversity, and when they are open to individuals from different backgrounds, these spaces do not feel safe. This makes the need for identity-related support evident. My talk highlights the importance of effective and empathetic mentorship and best mentorship practices for mentoring those with different identities from the mentee’s perspective. As a Black trans person, I was initially hesitant to be mentored by a white person because sharing one’s interests and admitting to not knowing where to begin felt vulnerable. During my talk, I plan to briefly share my story detailing how my mentor and I cultivated trust. This talk also provides solutions for growing diverse and sustainable teams where trust, learning, and experimentation are encouraged! I’ll pinpoint the strategies my mentor and I utilized over those six months and have continued to use to this day.
This talk is necessary for anyone interested in gaining more effective mentorship strategies as they guide individuals with different identities and for any mentees, like me, interested in advice on navigating mentor/mentee relationships. You don’t necessarily need to hold the title of “mentor” to utilize these strategies; most folks in this field guide others in some capacity. All in all, I aim to show how mentorship plays a crucial role in growing diverse and sustainable teams within the workplace.
Best practices and Learned Lessons in Refactoring Researcher-Developed Statistical R Packages
Naeem Khoshnevis and Mahmood Mohammadi Shad
We present two years of faced challenges and success stories of collaboration with biostatisticians. Research software engineering is a relatively new career field, and the structure, responsibilities, evaluation metrics, and collaboration boundaries are not firmly defined. All of these can build unreasonable expectations and can negatively affect both researchers and RSEs. The best practices and faced challenges discussed here can be an excellent heuristic for the current RSEs to improve productivity and give a better picture for the potential RSEs on what to expect. We discuss the followed software engineering best practices and learned lessons, including unit testing, functional testing, Continues Integration, documentation for users and developers, containerizing development environment, and practical use of version control systems. We also discuss requirements for addressing different types of projects based on the maturity of the research (completed, in progress, not started). Efficient communication is an essential part of the collaboration, which can include a broad spectrum of researchers (from undergraduate students to a well-established PI). We discuss the best practices for communication and how provide tips on how to prepare good documentation and test cases.
Our two years’ experience shows how taking care of small, yet important details can help researchers and RSEs in building sustainable and better scientific software.