Cloud computing is a growing technology that provides on demand shared pool of resources over the internet. Sharing of resources amongst the number of cloud users makes task scheduling a challenging issue. Task scheduling issue in many cases resolved by meta-heuristic approaches. This paper proposes a solution for task scheduling in a cloud computing environment based on the meta-heuristic, Genetic Algorithm. The proposed solution i.e. Modified Genetic algorithm (MGA) uses a hybrid solution based on Genetic Algorithm along with Predict Earliest Finish Time (PEFT) scheduling on Directed Acyclic Graph (DAG). Simulated results of the Modified Genetic Algorithm are compared with basic GA and with hybrid GA with HEFT (Heterogeneous Finish Time First) scheduling algorithms. Further, comparative analysis has been performed based on makespan, average processor utilization, processing cost metrics. It is observed that MGA gives optimal results in terms of processing cost and processor utilization for the unbounded number of processors.