Sediment erosion is one of the most notorious phenomenon damaging the turbines, in the Himalayan region. Francis turbine, which is one of the most used type of hydro turbine in the world as well as in Nepal, is affected quite severely, more so when operated in off design conditions. This study has taken a reference design of a Francis turbine from a hydropower in Nepal to optimize it for operation in variable speeds. Minimization of sediment erosion and maximization of efficiency are taken as the objective functions of the optimization, for which blade angles at trailing edge and blade angle distribution are taken as the design variables. The design space of the runner are constrained such that the optimized design could replace the existing runner in the turbine. Latin Hypercube Sampling technique is used to populate the design space such that the design variables are divided randomly to create required number of designs in the design space. Computational Fluid Dynamic analysis are performed on simplified numerical models of the samples to predict their performance and Sediment Erosion Rate Density (SERD), under various operating conditions. The results of output parameters, obtained from CFD, along with the design variables, are used to develop an approximation model relating the objective functions with the design parameters. NSGA-II optimization technique is used to search for the optimum design. The paper presents the comparison of the sediment erosion and efficiency of the reference runner and optimized runners under various operating conditions. It also presents an outline of the process used to optimize the runner for variable speed operation with minimum sediment erosion.