Risk, Controls & Governance
Ryan Miller’s executive project experience
Custody Declarations & Auditability
An operating-project case on a package-loss investigation that corrected its own headline figures and exposed gaps in declaration evidence. It connects event capture, physical verification, control workload and the test required before claiming recovered value.
Executive brief
Establish the complete declaration population before sizing or funding a new control.
Role
Ryan Miller · Prior executive role
Mandate
A complete-looking record can conceal an unobserved operating decision
Operating lens
Risk, Controls & Governance
Project overview
[AI] The investigation changes the investment question from choosing a loss-control threshold to establishing the decision and population that any threshold would govern.
Context
[HR] A logistics network investigated packages reported missing after recorded stow activity. The brief asked what happened between the recorded state and removal from a delivery itinerary. Its analytical corpus contained different populations, duplicate rows, evolving loss records and incomplete evidence about the declaring action.
Objective
[HR] Deliver a defensible explanation and a bounded operating ask. The recorded project owner constrained the work against unsupported theft assumptions and named employee rankings, required package-level traces and directed the correction of prior headline figures.
Approach
[HR] Machine-executed analysis compared entity grains, cohorts, event ordering, policy scope and earlier issue records. [AI] This case reconciles the register, tests the logic of its bounds and develops the Declaration-to-Outcome Evidence Contract (DOEC).
Findings
[AI] A distribution selected on subsequent loss cannot establish live control workload. Field completeness is conditional on emitted records; capture, joint completeness and truth require separate tests. Recorded scan states do not establish custody at the declaring moment. Different event anchors prevent direct pooling or cross-cohort ceilings.
Decision Value
[PD] Assign an accepting owner to the declaration contract, enumerate decisions independently and reconcile item identity across event versions. Calibrate capture and physical verification before a bounded response test. Retain current customer remedies and account for verification workload and displaced ordinary work.
Implications
[AI] The recorded contribution is diagnostic leadership and publication discipline. Implementation, accepted control ownership and realized savings remain unestablished. The proposed evaluation measures original-promise outcomes independently of how many incidents the new logging reveals.
Reading the evidence
The report distinguishes historical record [HR], recalculated findings [RC], analytical inference [AI] and proposed design [PD]. These labels describe the evidence supporting each statement about this actual project.
Contribution and evidence
What the record supports.
The analytical finding
Analytical finding: a loss-selected removal distribution cannot establish how often a live threshold fires. Likewise, complete fields among emitted events do not establish that all decisions were captured or that their contents are true.
Ryan’s documented contribution
Recorded contribution: owned the question, scope and deliverable; directed machine-executed analysis. Implementation remains unestablished.
Attribution
Work performed by Ryan Miller in a prior executive role at a global commerce and logistics company. Historical source artifacts were not recovered locally. This source-register reconstruction and proposed evaluation do not establish historical causal savings.
Evidence
Explore the evidence behind the work.
Custody Declarations & Auditability | 27-page public project report. Includes the executive storyline, technical methods, evidence boundaries and references.
Download the complete public case report (PDF)
The report protects internal identities and sensitive operating data. Private evidence annexes and source files are not published.