Secure multi-party computation (MPC) allows organizations to collaborate on sensitive data without revealing it. This opens the door to applications like joint fraud detection in finance, privacy-preserving medical research, and secure machine learning across institutions. By combining strong privacy guarantees with practical performance, MPC has the potential to reshape how data is shared and utilized across domains. My research advances this vision by developing protocols that are both efficient and resilient to active attacks and leakage. These include constant-communication garbled circuits, round-optimal MPC, and large-scale deployments designed to bridge the gap between theory and practice.