I am currently working as a Software Engineer for the MeerTRAP project at The University of Manchester. Here, I work on the development of pipelines, often venturing into other areas such as Machine Learning classification of our data, web development, system administration and processing techniques for time-domain signals. I am also responsible for the development and maintenance of our processing and post-processing Docker images, where I am always trying to build smaller, faster and less complex images.
My research interests are mainly focused around the use of High Performance Computing for the astronomical transients search pipelines. That involves initial stages of processing, such as signal processing as well as post-processing including machine learning and the presentation of the final research products to the users. I am also interested in developing better software with the help of containers, where I am mainly researching the efficient use of Docker and Singularity-based research software stacks.Profile Picture – adass-xxxi-2021/question_uploads/1352d021-92ad-4890-9871-bae75ea9e716-5173_4dZw1Bw.jpeg Affiliation –
The University of ManchesterPosition –
Software EngineerTwitter handle –
fastradiopotatoHomepage – GitHub ID –
mmalentaPostal address –
Alan Turing Building, 3.223
Department of Physics and Astronomy
The University of Manchester
Modern research is already highly reliable on containers and this reliability is only going to increase over the coming years. Some of the most common software stacks are already distributed in their containerised form and many more are soon going to join them.
The use of containerised research software is especially important in the context of reproducible research. Here containers can be used to create and distribute consistent research software environments. They have their uses in complex installations of interconnected libraries as well as relatively simple requirements, such as stable python environments. Containers also make the software stack and research much more accessible, by removing the need for often time-consuming installation processes and providing an isolated computing ecosystem that can be treated like a black box where only input data has to be provided by the user.
Despite their widespread use, many misconceptions and bad practices are currently prevalent within the research community about the abilities, use cases and safety of containers. If used incorrectly, containers can produce inconsistent results, fail to deploy or even expose critical security vulnerabilities to potential attackers. It is therefore important to equip the scientists with skills that will let them produce and distribute containers that are efficient, easy to use and safe from the point of view of the user and the developer.