 Hi, my name is Xin Wang, and I work at Intel. Today, I'm going to talk about how to build trusted cloud applications with the Web Assembly on Intel SGX. The presentation contains a few use cases provided by Michael Bowman from Intel and Jia Zhang from Alibaba. Building a trusted cloud application is not an easy job if you are the application owner. You probably concern how to protect your cloud code and the data from being accessed by either the CSP, hackers, or other malicious cloud applications. For the CSP, they need to ensure that infrastructure can resist the attacks from any malicious cloud applications. They also need to prove to their customers their concerns can be solved technically. And also, in some cloud applications, it will consume the data from third party. We also need to address the data owner's concerns for data leaking or misusing. It is hard to address all the challenges by pure software solution. Typically, we need to leverage some hardware capability, such as trusted execution environments, or we call it TEE. The TEE provides hardware-based isolation for the code execution and the data. TEE usually provides remote attestation capability that is also very needed for building trusted cloud applications. Intel SGX is the most researched, updated, and battle-tested hardware-based TEE for the data center. It's available across the entire line of isolate scalable processes. It delivers the smallest potential attack surface of any TEE available for the data center. On the SLAIC Xeon processor, the SGX supports up to one terabyte protected enclaves of data in the code. WebAssembly provides a lightweight sandbox that supports scalable concurrent execution of different Watson modules. By using Watson on SGX, we can leverage SGX to protect the applications from external attacks and also leverage the Watson sandbox to protect the host environment from internal attacks by malicious applications. WebAssembly makes a long time. Or Walmart is an open-source project on the bank code alliance with supporting SGX as the first class citizen. Walmart provides a full feature set for SGX support. Firstly, the Walmart is designed for small TCB. The total binary size can be 100 kilobytes. And the Walmart supports interpreter and the ahead-of-time compilation for the Watson execution on SGX enclave. The Liberty is also supported for the Watson application based on Intel SGX SDK. The Watson symbi is also supported by Walmart on the SGX. We have built some IA framework such as TensorFlow to Watson and successfully executed from SGX enclave. So far, Walmart has a good adoption progress by either open-source or commercial usage with SGX. Now I will get to introduction for a few solutions that is Walmart on SGX. Private Data Objects is an open-source project on Hyperledger Labs that enables sharing of data and the co-dilating amount mutual distribution parties. PDO provides a secure framework for loading Watson-based smart contracts into SGX enclave and the executed with WebAssembly micro runtime. Enclave container is another open-source project that uses Watson on SGX. The Enclave project is an infrastructure that builds confidential containers for TEE and extends Kubernetes across different TEEs. As the diagram illustrates, the Enclave container can build the Watson modules and the Walmart runtime into a confidential container with some new plugin and components that they introduced into this project. The confidential containers can be orchestrated by the Kubernetes across the TEE. This is a great project that solves the challenges without application modifications. The last use case here is a usage of using Watson as SGX for multi-party computation. In this use case, each party develops a Watson application that can be either a consumer or a supplier or in both roles. All the parties review the source codes of WebAssembly applications and assign the binary modules. Our base infrastructure will load the Watson modules into SGX enclave and the support, establishing secure and trustworthy channel between the Watson modules. In that way, the multi-parties can share the data for computation safely. I have finished the presentation. And thank you for listening. If you are interested, please go to the GitHub and download Walmart and try it out with SGX. Thanks.