Rooted in Approximation Theory, Optimal Recovery is a trustworthy learning theory focusing on the worst case. Regrettably, compared to more popular Machine Learning alternatives, the classical theory of Optimal Recovery neglected the computational aspect, with a few exceptions, e.g. the development of spline functions. Nowadays, modern optimization techniques facilitate advances---even theoretical ones---on the minimax problems that abound in the field. I will illustrate this point by selecting a few snippets from my recent work: 1. Prediction of (multivalued) functions based on merely convex models; 2. Estimation of linear functionals in space of continuous functions; 3. Full Recovery from deterministically inaccurate data in Hilbert spaces; 4. Estimation of linear functionals from stochastically inaccurate data; 5. Prediction of the maxima of Lipschitz functions from inaccurate point values.