We report about ongoing work developing infrastructure and libraries for data-analysis in the newly released, dependently-typed programming language Idris 2. Dependent-types allow us to express nuanced dependencies statically, and we want to explore to what extent they can help with data organisation, cleaning, modelling and analysis, and visualisation. Over the last couple of years, we developed infrastructure and libraries supporting activities along the data-analysis pipeline. These include: dependently-typed regular-expression parsing for data-cleaning and extraction; tabular data interfaces for manipulating and reshaping data-frames; modular probabilistic modelling libraries and modular implementations of common inference algorithms; and bindings to popular data-visualisation and storage formats to emit results and reports. We will also present ongoing work involving using these libraries for a probabilistic generative music library.