We evaluated the effects of a 10-week intervention that randomly provided access to remote training to parents of preschool children during summer vacations in Peru. In response to learning losses during COVID-19-induced school closures, educational coaches offered guidance and support to parents in activities designed to accelerate the development of foundational math skills. We found that the intervention improved cognitive scores in mathematics by 0.12 standard deviations. Furthermore, we show that remote trainers increase the likelihood and frequency of parental involvement in math-related activities, suggesting that improvements in learning are driven by greater parental involvement in children's skill development.