AI/ML is the most important data-driven application. In our team, we focus on solving some of the biggest open-problems in AI security and safety, including

• The certified robustness of various AI systems: we’ve worked on understanding the fundementals of AI safety and security in open-world and adversarial deployment.

• Data security and privacy in AI training, inference and model-service: we’ve worked on various techniques (such as privacy-preserving machine learning, federated learning, ZKP) to ensure data security and privacy in AI.

• Intelligent Network DataPlane (INDP): Our team is the world leader in embracing the flexibility of programmable dataplane (e.g., programmable switches, SmartNICs, and host networking) to execute application-desired and data-driven traffic shaping/analysis logic, while imposing negligible overhead on the network data plane.

• Evolving the control-plane, data-plane, and management-plane for production-grade datacenter and WAN networks to enable various high-performance workloads, such as LLM training across large-scale data centers.

Blockchain is the fundamental infrastructure of the next-generation trust-based Web (the so-called Web3.0). Web3.0 is often cited to shape every aspect of our lives:

• Build transparent and trustworthy computing and ground-breaking Web3.0/Metaverse applications, such as transforming the legal sector via smart contract.

• Build security and trust systems to empower Internet architectural initiatives, such as verifiable routing and IP network.

• Privacy Computing: we worked on various privacy computing techniques, including Federated Learning, Data Space, Secure Multi-party Computation, Secure Enclaves (i.e., trusted execution environments), Zero-knowledge Proofs, and Differential Privacy systems.

• AI-Driven Network Traffic Analysis: networking security heavily depends on the insights into encrypted traffic. Our team built various algorithms and systems to enable accurate analysis.

• Secure Inter-domain Routing: our team is leading in building a secure inter-domain routing protocol for the Internet.

• Readily deployable and proactive DDoS prevention systems: CCS’16, ToN’18, TIFS’18, U.S. National NSF I-Corps Award.

• Privacy Preserving access control for Tor Networks: ICNP’17. Acknowledge by the Tor Project and Cloudflare Inc.

• Internet Source and Path Authentication: IWQoS’18.

• World’s first production-grade SDN control-plane: NDSI’21 (Special acknowledgement), Google Feats of Engineering Award for High Network Availability.

• World’s first production-grade automated management plane: leading author and under review, Google Feats of Engineering Award for WAN Capacity Augment.

• Managing multi-tenancy and routing control in public Cloud: INFOCOM’18, TPDS’19, TPDS’21, One U.S. Patent, Product adoption by NEC, Acknowledgement by Huawei and Google.

• Pioneering work in Blockchain Interoperability and Web3.0 definitions: CCS’19, TDSC’21, Invited Talks @ IEEE Blockchain Standards, Ant Group, IC3, etc.

• Blockchain-empowered application innovations: IEEE Network’20, ICDCS’19.

• Vulnerability discovery for large-scale RPC systems: ASIACCS’21, Adoption by Ant Group.

• Advanced Persistent Threat: CCS’19.

• Searchable Symmetric Encryption: TIFS’20.