TECHNICAL PROGRAM LEADERSHIP

Program delivery.
Evidence-bound AI.

From engineering programs and customer delivery to AI-assisted workflow design: define the work, make the acceptance criteria explicit, and show what the evidence actually supports.

This portfolio presents three concrete examples, each with an accessible artifact and a clear boundary around the claim.

OPEN EVIDENCE PACK · PDF APPROVED RÉSUMÉ · PDF

CASE 01 · PROGRAM DELIVERY

Connect contract scope to accountable delivery.

Customer-facing services and engineering programs require more than scheduling. They require alignment between commitments, delivery practices, reviews, and acceptance.

William’s contribution

At Siemens Digital Industries Software, led customer-facing services work involving contracts, statements of work, delivery alignment, and business practices. Earlier NAVAIR/NCMS responsibilities covered concurrent engineering programs and delivery accountability.

The evidence boundary

The figures shown here come from the approved résumé. The recurring inefficiency estimate is an identified opportunity, not a claim of realized savings. Employer records and client deliverables are not published.

CAREER DELIVERY BRIEF · PDF

Source: William Earl Lemon’s approved résumé, Siemens (2021–2025) and NAVAIR/NCMS (2017–2021).

CASE 02 · AI GOVERNANCE ENGINEERING

A recommendation is not an authorization.

The CIL work addresses a specific contract problem: a model review, deterministic eligibility decision, human-readable card, and downstream sizing proposal must not silently become different versions of the same decision.

William directed the requirements and review boundaries for an AI-assisted implementation. The delivered slice is offline Python contract code, not an activated trading system.

CIL reviewBounded review inputs and required-member evidence.

EvaluatorExplicit policy, facts, risk inputs, and refusal conditions.

Decision cardTen answers projected from the same decision record.

PEARL payloadIdentity-bound handoff. Orders remain disabled.

Implemented in the bounded slice

Missing observations or required review evidence produce a hold.
Decision records bind the policy, evidence, risk inputs, and ten answers.
Changed decision content fails identity verification.
The handoff carries disabled activation and order flags.

OFFLINE TEST RECEIPT

89 passedTwo named contract-test modules, rerun September 16, 2026.

Synthetic inputs, network-free test scope. This count excludes the legacy platform suite and all broker, feed, model-provider, database, and deployed-workflow behavior.

CODE EXCERPTS + TEST RECEIPT · PDF ORIGINAL CHANGE RECORD (access note)

Implementation: policy evaluator (access note) and decision contract (access note). Public PDF excerpts are available without a GitHub account; repository access may vary.

THE CONTROL HAS TO SAY NO

Success and refusal, side by side.

These are outputs from synthetic fixture executions, not market observations or investment recommendations. A complete example qualifies only for paper review.

Observed offline fixture outcomes on September 16, 2026
Test input	Observed result	Meaning
Complete synthetic LONG and SHORT	`ELIGIBLE_FOR_PAPER_REVIEW`	Review eligibility, not permission to trade.
Ten answers omitted	`DEFERRED`	The card cannot bypass missing answers.
Required reviewer or observations absent	`HOLD_FOR_EVIDENCE`	No invented review completion or facts.
Decision content altered	`verification = False`	The changed record fails its identity check.
Constructed PEARL handoff	`order_enabled = False runtime_activation_allowed = False`	No execution path is enabled by this payload.

An identity hash is not proof that a fact is true, a database write succeeded, or a broker accepted an order. Those require separate trusted evidence and integration tests.

Source: CIL Decision Contract Evidence, generated test receipt and original excerpts from the bounded fixture suite (access note).

CASE 03 · RELEASE VERIFICATION

Make acceptance criteria observable.

The public site needed readable text, stable logo alignment, a wider access form, and the approved résumé presented as PDF only. A rejected visual change was rolled back rather than defended.

The measured release

Browser checks recorded 52 page/viewport observations across 26 HTML pages at 375px and 1440px widths: no measured horizontal overflow and a 14px minimum visible text size.

Targeted visual checks

The Gate and Request eye were measured at 160px, centered on the upper logo circle before and after scrolling at three widths. The request form has a 640px desktop maximum width.

These are bounded UI checks, not a complete accessibility audit. They do not establish email delivery, approval notifications, authenticated behavior, or Telegram integration.

RELEASE VERIFICATION RECEIPT · PDF

Release history: preserved rollback, PR #12 (access note) and accepted readability update, PR #13 (access note). Measurements and exclusions are reproduced in the PDF.

Actual local browser capture from the September 16 readability checks. The remote status endpoint was stubbed offline during this test; the status label is not production telemetry.

MEASURED UIROLLBACK PRESERVED

OPEN THE WORK

Artifacts, without an approval queue.

These public PDFs open directly. The briefs are newly assembled portfolio summaries; the contract packet also reproduces original code excerpts and actual test output. Private working records and credentials are not included.

ADDITIONAL WORK · FREE USER-OPERATED SOFTWARE

AI Model Release Monitor

An MIT-licensed interface for evidence-linked AI releases, selected workflow-fit reviews, private in-page stack matching, and documented version/access comparisons. Run it locally or on your own hosting; separate collection services and infrastructure are the operator's responsibility.

Updated September 21, 2026. Coverage is partial; cost scenarios are not measured savings. No managed update service is included. This project is separate from the three PDF case studies above.

GET SOURCE & INSTRUCTIONS SCOPE & SOURCE

COMPLETE REVIEW PACK

Portfolio Evidence Pack

All three case-study artifacts in one PDF, with source URLs and explicit limitations. The approved résumé is a separate download.

OPEN COMBINED PDF

CAREER · RÉSUMÉ-SOURCED

Career Delivery Evidence Brief

Responsibilities, program scope, and identified improvement opportunities. Self-reported career evidence, not an independent employer audit.

OPEN CAREER PDF

ENGINEERING · OFFLINE TESTED

CIL Decision Contract Evidence

Implementation boundary, 89-test receipt, positive and refusal outputs, and original contract/test excerpts.

OPEN CONTRACT PDF

DELIVERY · BROWSER-MEASURED

Release Verification Receipt

Page/viewport results, eye-alignment checks, screenshots, and the change-and-rollback record.

OPEN RELEASE PDF

MATURITY, WITHOUT INFLATION

What this portfolio does and does not establish.

Available to inspect now

Approved résumé and attributed career examples.
Offline CIL contract implementation and bounded test results.
Published site release history and measured UI receipts.
Direct, public PDF evidence artifacts.

Not claimed by this evidence

A complete or operationally proven AI estate.
Live trading, broker execution, or investment returns.
Verified employer savings or employer endorsement.
Proven email delivery or automated access approvals.

Other estate components remain separately scoped in the Registry. A completed portfolio release does not change their implementation or validation status.

START WITH A REAL DELIVERY PROBLEM

Technical leadership that makes the handoff inspectable.

The strongest fit is work connecting requirements, delivery execution, AI-assisted evaluation, and acceptance evidence. For a role, pilot review, or development-support conversation, the artifacts above provide a concrete starting point.

TECHNICAL PROGRAM LEADERSHIPAI EVALUATION & GOVERNANCEDELIVERY ASSURANCE

EMAIL WILLIAM VIEW RÉSUMÉ & CONTACT LINKS RÉSUMÉ PDF

Public contact: ashes2echoes.platform@gmail.com. Additional private material remains subject to an explicit access grant.