Good scientific methods need to hold up against messy data and real working conditions. I approach each project through these operational principles:
Start with the question, the data, and the tools. I examine how data were produced, what they can support, and define feasible operational targets, given data, resources and methods available.Â
Make the work reproducible, expand from a minimum viable product. I keep data transformations, analytical choices and quality checks reproducible. I start with a minimum viable product and expand and re-run protocols and analytical pipelines to obtain more refined and detailed results, as agreed with the client.
Take uncertainty seriously. Missing data, measurement error and consequential assumptions should be part of the analysis, and key for a proper risk/reward assessment.
Design for conditions on the ground. A method must be feasible for the people who collect, manage and use the data, within their time, resources and institutional constraints. Those conditions should shape the method from the start.
Connect people and scales. Field teams, local partners, collection staff and researchers often see different parts of the same problem. I bring those perspectives into scientific and technical decisions, with attention to local priorities, clear roles and appropriate credit.