Decentralized applications built using smart contracts are skyrocketing. These applications take advantage of blockchain’s availability and security guarantees. However, blockchains have failed to cope with the increase in adoption because of inherent limited scalability and poor throughput. This prevents mainstream adoption of application execution on blockchain. In this paper, we present SLAB, a novel smart contract architecture that addresses these limitations by proposing parallel smart contract execution. We separate computation from consensus layer and leverage Intel SGX to scale transaction execution. SLAB uses three major concepts for optimization. First, we use smart locks and dependency tree to enable transaction execution in parallel. Second, by utilizing the trust properties of Trusted Execution Environments (TEEs), SLAB eliminates redundant transaction execution. Third, we support complex smart contract to smart contract calls by smart allocation of transactions and communication between compute nodes. We claim that our system is highly scalable. And, as compared to ethereum, we expect a many-fold increase in system throughput.