Wei Dai

I am a researcher/developer specializing in privacy-enhancing technologies and responsible AI. I am enthusiastic about applying research advances on products and services to protect data privacy and enable collaboration. And I care about usability, maintainability, and security in software development. In a broader scope, I investigate privacy vulnerabilities in products and services from end to end. I am currently investigating privacy, security, and robustness in AI systems and aiming to find formal definitions. I also actively engage in discussions with production and legal teams related to policy-compliant data sharing and collaboration. Across all of these project areas, I act cross-disciplinarily, partner with internal or external collaborators, mentor interns, perform independent development work, and present results to customer, academic, industry, and government audiences. I actively serve on program committees, review peers' works, and publish research papers. I have released and contributed to many open source projects. Before joining TikTok, I worked at Microsoft Research. I led the cross-architecture, cross-platform development of Microsoft SEAL homomorphic encryption library and its downstream products and services. I explored compatibility on embedded devices and acceleration on GPUs and FPGAs. I advised Microsoft engineers to build software support for Intel's ongoing ASIC design funded by DARPA DPRIVE. I built usable automation tools and production-ready secure protocols.