Recombinant protein expression is central to academic exploration as well as biotechnology’s advancement of human health, climate applications and the bioeconomy in general. However, not all proteins can be expressed in all organisms, and the field lacks a predictive model of soluble protein expression that could replace laborious experimental trial-and-error. This project aims to design and test an extensible experimental platform and standardized data ontology for collecting soluble recombinant protein expression data across organisms. The resulting dataset will be used in building increasingly generalizable predictive models of protein expression. In this document, we have roadmapped data collection techniques, designed experiments to assess their feasibility, and outlined a proof-of-concept pilot study that would be conducted prior to full-scale data acquisition. Here, we propose a plan of action that will establish the feasibility of gathering ML-ready soluble protein expression data in two organisms commonly-used in biomanufacturing and protein expression: Escherichia coli and Pichia pastoris. Results of the feasibility study will inform decisions about the specifics of future pilot and full-scale data acquisition. We encourage the reader to reference our review, “Can protein expression be ‘solved’?” for details on soluble protein expression. Once gathered, the dataset proposed here will adhere to FAIR principals, be publically available, and “living” – researchers can freely use it and it will continue to grow over time. Subsequent experiments (beyond those outlined in this proposal) will include examining more protein sequences and expression in different organisms (e.g. Bacillus subtilis, Aspergillus niger) to make the model more extensible.