The ability to solve complex realistic problems is the most important requirement for being a good scientist or engineer. Here we present a theoretical design for improving the effectiveness and efficiency in teaching technical problem solving at the post-secondary level. The approach brings together theoretical ideas and empirical results from (a) evidence-based instructional practices, (b) development of expertise through deliberate practice, and (c) a decision-based framework for complex problem solving. This design involves explicit practice and feedback on making the set of decisions that define the problem-solving process of skilled scientists and engineers. We provide a template for instruction based on this design. This design also provides a way to assess problem-solving skills that is more accurate than traditional examinations.