This paper proposes a public auditing mechanism called Panda that allows for efficient user revocation in shared data stored in the cloud. Existing mechanisms require re-signing data blocks signed by a revoked user, which is inefficient. Panda utilizes proxy re-signatures to allow the cloud to re-sign these blocks on behalf of existing users, improving efficiency and reducing users' computation and communication costs. A public verifier can still audit integrity of shared data, even if parts were re-signed by the cloud. Panda also supports batch auditing of multiple tasks and can scale to large user groups and data sizes.