Release Train Engineer, Data 1^st Project

The Data 1st Project was strategic to consolidate and improve the company’s Data quality. We started with very urgent use cases, to then establish a standard methodology to be transferred to business as usual later on.

The biggest challenge for the Project was to extract data from multiple different source systems, transform and standardize them and load them into a Data Lake. Once there, the Data Science team would analyse them and build tools for the business to use in decision making. In parallel, the Project also was building a single consolidated Product Taxonomy data base, so later could be use within the tools built by the Data Science team.

I was invited to be the RTE temporarily, until they found a full-time employee that could do the role. Even though the Programme had already 3 Programme Increments of SAFe experience (about 9 months), there was still much to improve.

On one hand, we change the team structure. We built Feature Teams, so each of them could deliver value independently of the others). Before this change, any single requirement, would require the collaboration of at least 4 different teams to be delivered. This would lead to delays and continuous prioritization discussions.

On the other, we spend a lot of effort establishing a Feature Preparation Process for the next PI Planning. Experience had shown us that during the last few PI Plannings, the teams spent most of the time discussing requirements and agreeing Acceptance Criteria, rather than actual detail planning. The root cause of this was clearly that Features reaching PI Planning had not met the “Definition of Ready” yet.

And lastly, we also put emphasis in defining, demonstrating and visualizing the Continuous Deliver Pipeline, or the End to End flow of Value. How each team or skill within the Programme was fundamental to the success of it, but that they we not independent, but rather they should all work collaboratively to in order for the Programme to deliver value periodically.