Back of the envelope calculation for machine learning projects

One of the best practices that we know from great engineers is the back-of-the-envelope calculation to estimate costs and resources. I believe that in Machine Learning Engineering, we all would benefit from such a “back-of-the-envelope calculation” skill to create a prototype of an ML Project. We need to confirm - as cheaply as possible - that our future ML project is worthwhile, that it will solve my business problem, and that costs and resources are feasible. In my talk, I suggest a design toolkit for ML projects to perform such rough prototyping by using three canvases: Machine Learning Canvas, Data Landscape Canvas, and MLOps Stack Canvas.