Abstract
Ensuring reproducibility and trustworthiness in intelligent systems has become a cornerstone of applied systems innovation. This study introduces HyCoSS++, a deterministic machine learning framework with simulated ledger-based audit anchoring that formalizes verifiable reproducibility and auditability within AI-driven architectures. The proposed approach integrates cryptographic provenance mechanisms - SHA-256 hashing and Merkle root aggregation into a calibrated ensemble learning pipeline designed for consistent, transparent, and tamper-resistant computation. By embedding deterministic execution and version-controlled environments, HyCoSS++ guarantees identical outputs under identical conditions, addressing the reproducibility crisis across computational sciences. Beyond algorithmic performance, the framework represents a system-level innovation, combining reproducible computing, trustworthy AI, and cryptographically verifiable auditability to create verifiable intelligent infrastructures. Experimental evaluation demonstrates stable calibration, traceable model lineage, and high reproducibility without compromising accuracy. The proposed methodology provides a transferable foundation for auditable, explainable, and reproducible intelligent systems, aligning with the core mission of applied system innovation research.