Tu sei qui

Reproducible Science: a dev ops take on science and research

21 novembre 2019
3:00 pm
San Francesco Complex - Classroom 2

Scientific results are the pillars of the next wave of results.

It is imperative to be able to build on solid foundations and able to share finding that others can leverage and observe.

The challenge of creating and sharing material is not a simple one, and often results are not easy to reproduce, this is a crucial issue when other wants to leverage current research results to build their theories. Lack of regression testing on experiments, version management and cooperation methodologies brings risk to the credibility and impact of research work.

The DevOps world had to face such issues for a long time in production environments and at a large scale, lots of different tooling flavours but always a single goal in mind: capitalise on previous assets, version management for correct consumption and regression testing for early catching shifting behaviours.

Adopting and adapting such methods would improve the research processes; such endeavour is a significant core pillar of the Alan Turing Way promoted by the Alan Turing Institute and applies to a wide plethora of disciplines.

In this session, we will cover scenarios and approaches that could lead to the production of “reproducible science”.

Diego Colombo, Microsoft UK