The proliferation of autonomous systems, ranging from advanced robotics to high-stakesalgorithmic trading and Artificial General Intelligence (AGI) substrates, has precipitated a crisisin systems software reliability. Traditional languages, designed in an era prioritizing executionspeed over correctness, expose critical infrastructure to non-deterministic failures and undefinedbehaviors. This paper presents a comprehensive, architectural analysis of Aria, a novel systemsprogramming language designed to enforce a "Safety Above All Else" philosophy. Through arigorous examination of the compiler source code, runtime specifications, and theoreticalunderpinnings, we dissect Aria’s radical departures from conventional language design. Weanalyze the Twisted Balanced Binary (TBB) type system, which enforces symmetric arithmeticwith embedded error sentinels to prevent silent overflow corruption. We explore the Limb-BasedIntegral Model (LBIM), a structural workaround for backend optimization instability that enablessafe post-quantum cryptography. Furthermore, we investigate the "Sentinel DiscontinuityConstraint," the "Appendage Theory" of metadata management, and the language’s dual-memorymodel (GC/Wild) utilizing the Acquire-Convert-Release (ACR) pattern. By synthesizing detailsfrom the compiler's Intermediate Representation (IR) generation and its runtime environment,this report demonstrates how Aria achieves bit-identical determinism and sticky errorpropagation, establishing a new theoretical and practical framework for high-integritycomputing.