The ability to preserve fragile quantum information for extended periods is a foundational requirement for practical quantum computing. This project develops a new class of ultra-high-performance quantum memories by systematically exploiting symmetries, underlying patterns that govern how physical systems behave, to enhance the protection of quantum data. In doing so, it supports the progress of science by advancing core knowledge at the intersection of quantum information theory, physics, and machine learning. The results of this research may accelerate the realization of large-scale quantum computers, enabling breakthroughs across areas such as materials design, secure communication, and artificial intelligence. The project also contributes to the national interest by strengthening U.S. leadership in quantum technologies and preparing a diverse future workforce through interdisciplinary training of students and postdocs. Technically, the project integrates symmetry principles into all aspects of quantum memory design and decoding through three coordinated thrusts. The first develops capacity-achieving low-density parity-check (LDPC) codes for biased noise using emergent symmetries generated via Clifford transformations, enabling more efficient and accurate decoding. The second investigates a novel route to self-correcting quantum memories by constructing stabilizer codes in three dimensions that display symmetric properties between different types of quantum errors, offe