SELECTED WORK / INTERACTIVE EDITION
Operate the system.
Inspect the evidence.
What does the evidence support? What happens when delivery fails? Who checks the AI?
Three ways to explore how I think about systems people depend on.
01 / CLAIMS VS. REALITY
One number. Two interpretations.
The displayed value invites a battery-life conclusion.
THE CLAIM TO EXAMINE
“The battery died early.”
Reconstructed display based on the historical case title. Not a captured device screen.
BEFORE THE CONCLUSION
What would you need to know?
Loss of contact, loss of power and degraded battery capacity are different claims. Start by separating them.
- Did transmission stop?
- Did the device keep recording?
- Was battery capacity actually measured?
OPEN THE SOURCE / SECONDARY NARRATIVE
The limitation is part of the finding.
Exact excerpt from the historical audit narrative
Alternative explanations and limits. Parameter 102 read 0 (sleep disabled); recorded cadence changed from 50.5 to 55.4 seconds. Charging voltages of 4,069–4,125 mV do not establish battery capacity or rule out degradation. Temperatures ranged from 27 to 43 °C; this sample alone does not rule out temperature effects. Controlled discharge testing would be needed to separate these causes.
This is the source text used in this portfolio, not a device log or independent measurement. Its charging-voltage observation does not rule out battery degradation.
Source provenance
Document: sample-audit.md · Read the rendered document
SHA-256 identifies the exact source version used for this page:
de335ff76d4c6ddf6f762ffe2e963b408a8c82267a9d80833eb9add00876736fThe existing DiTrack audit narrative reports transmission stopping at 3,596–3,632 mV while buffering continued. Raw device logs have not been independently re-run for this portfolio.
Read the full audit ↗DITRACK / ONE DEPARTURE, TWO DIFFERENT OUTCOMES
The truck left.
Did the alert arrive?
Follow one truck out of its depot. Then see why detecting a departure and delivering its alert are two separate responsibilities.
Interactive explanation · synthetic data · no messages sent- 1At the depot
- 2Cross the boundary
- 3Keep tracking
- 4Check the outcome
Start with a simple rule.
When truck 01 leaves this depot, notify the recipient. The truck is still inside, so there is no departure to report.
Inspect what was recorded ↓
These records belong to this demonstration. Coordinates are positions on the illustration, not GPS coordinates. Step numbers show sequence, not elapsed time.
| Step | Record | Detail |
|---|
Keep vehicle processing independent of notification delivery, and retain records that make failures traceable. This original illustration is informed by reviewed DiTrack code for jobs, notification services and delivery records. The outage, recovery and route are scripted teaching examples; they do not establish production timings, retry guarantees or the current deployment.
Engineering background (PDF) ↓ANONYMIZED AI EXPERIENCE / ORIGINAL DEMONSTRATION
A draft is a hypothesis.
Put it through a check.
Find every real solution. The draft below has lost one.
01 / WORKING DRAFT
Edit the values yourself. Try 0, 1, then add 2. These checks validate this equation’s solution set, not arbitrary prose.
02 / CHECK RESULTS
Not checked yet.
03 / REVIEW HISTORY
- v1 · Original scripted draft loaded. No checks completed.
This original demonstration illustrates review practices informed by my educational-content work. No company, proprietary architecture or internal product screen is reproduced. No live model is called. Passing the numeric checks does not certify the written explanation; human review is required.
FROM THE EXAMPLE TO YOUR SYSTEM