Arya forge audits and approves financial data releases securely
Arya Forge: Auditing Validation and Release Approval for Financial Instruction Data
Artificial IntelligenceComputers and SocietySoftware Engineering
Summary
Financial data systems need to check not only if data is valid but also if it's okay to release it. The authors created Arya Forge, a system that links content checks and approvals to specific data sets. They tested it with many examples, showing it catches data mistakes and policy issues while keeping most good data. The system also tracks how costly processing is and how to recover if needed. Their work shows there is a balance between strict data rules and keeping useful data formats.
What this means in practice
- •For financial data engineers: Detect and reject faulty or unauthorized financial data releases by linking validation and approval directly to data manifests.
- •For data governance teams: Improve auditability and policy enforcement by applying approval coverage bounds and integrity checks within financial instruction pipelines.
Authors
Kushagra Bhatnagar, Deekshith Marla
Abstract
Financial instruction-data pipelines must distinguish content validity from permission to release a particular dataset. Arya Forge combines typed records, current-content checks, and approval coverage bound to a dataset manifest. We evaluate its release boundaries in a financial-data systems case study. An integrity audit covers 100,000 synthetic instruction records; a separate arithmetic implementation agrees with all 15,000 numeric records across twelve formula families. In 240 scenarios with 2,400 treatment rows, adding financial and release-record checks to a short-instruction filtering recipe rejects all injected numeric faults and policy holds while retaining 220 of 240 clean controls. The twenty lost controls comprise every tested cash-conversion-cycle example. Nineteen persisted-release cases show that eleven selected negative cases are blocked, four equivalent representations are rejected, and changes to external evidence after binding are outside ordinary Gold promotion's checks. Local processing measurements and controlled checkpoint resumptions characterize cost and recovery through 100,000 input rows. The results expose a tradeoff between strict producer-contract enforcement and useful-representation retention, and identify the protected inputs required for manifest-bound approval. These findings apply to the tested implementation and workloads; they do not establish general financial correctness, authenticated approval, expert-validated quality, or downstream model improvement.