# AI Deployment Context — Jermaine E. Whiteside Evidence CV **Context ID:** JW-CV-AIC-001 **Context Version:** 1.0 **Release:** v2.7 Candidate 47 **Date:** October 18, 2026 **Status:** Governed deployment handoff / cold-start reconstruction artifact ## Purpose This file is the authoritative AI handoff artifact for this deployment package. It is designed to let a capable AI system reconstruct the relevant deployment state without relying on prior conversation memory. It is not a transcript dump and must not be treated as proof of facts that are not independently supported by the package. ## Read-First Instruction to Any AI Before modifying this package, inspect this file, `AI_DEPLOYMENT_CONTEXT.json`, the current release notes, the current deployment manifest, `/deployments/registry.json`, `/applied/registry.json`, `/evidence-cv/registry.json`, and the relevant human-facing pages. Preserve the existing architecture unless the requested change explicitly requires alteration. Do not infer missing facts, deployment targets, certification status, or external-integration readiness. ## Canonical Production State - Canonical production site: `https://cv.jermainewhiteside.com` - Current deployment method: static Cloudflare production workflow. - Production must remain operable without ORCID/API runtime. - Production project/route history has previously diverged from the beta Pages project; a successful upload is not proof of production promotion. - After every deployment, verify the custom production domain and the visible version footer. ## Integration Laboratory - Beta laboratory: `https://beta.jermainewhiteside.com` - Beta project: `jw-certified-cv` - Purpose: ORCID/API and later scholarly-integration testing. - Rule: no beta integration is promoted into production until it passes documented verification and rollback testing. ## Current Rollback Baseline - Immediate rollback baseline: v2.5 Candidate 15 static production package. - Candidate 14 is the prior static recovery baseline. - Candidate 13 is preserved as evidence of a split-target deployment attempt and must not be treated as the production baseline. ## Locked Governance Rules 1. Production must never depend on an unverified integration. 2. Test → document → certify → promote. 3. Social media is a derivative dissemination layer; the CV Applied Evidence record remains canonical. 4. Controlled working records must not be silently promoted into certified CV claims. 5. Human-readable and machine-readable representations should describe the same governed structure. 6. Deployment history, version footer, release notes, and SHA-256 manifest are mandatory release controls. 7. Preserve provenance and rollback information; never overwrite prior deployment history. 8. AI continuity should be reconstructible from governed external artifacts, not dependent on model/account/session memory. ## Current Applied Evidence State - AE-2026-001: Faithful Intelligence — LinkedIn native-video derivative recorded. - AE-2026-002: AI Companion Technical Architecture — LinkedIn native-video derivative recorded. - AE-2026-003: 2040 South Church-derived applied-experience development queue. This is a controlled development record, not a certified CV revision. ## Candidate 16 Objectives - Replace the raw-JSON-only “Open Master Registry” experience with a human-readable Master Registry page. - Synchronize the Evidence CV master registry with the current governed evidence architecture, including Applied Evidence as a related governed registry. - Preserve direct access to raw JSON as the machine-readable representation. - Introduce this AI Deployment Context in both Markdown and JSON. - Register the AI context in the site machine manifest and deployment history. - Preserve the static production architecture and beta/API separation. ## Known Research/Test Program The deployment package is also a test artifact for a proposed AI continuity methodology. Planned validation includes: - repeated use across existing institutional websites; - a cold-start ChatGPT account with memory disabled; - a separate Claude account receiving the same frozen control package and minimal task instruction; - identical acceptance criteria across models; - later cross-model handoff tests. Primary measures under development: context reconstruction accuracy, constraint preservation rate, unprompted error rate, clarification burden, human intervention required, deployment fidelity, evidence-governance fidelity, and turns/time to a valid artifact. ## Research and IP Boundary The continuity architecture may generate scholarly and intellectual-property questions, including potentially protectable implementation features. Do not represent any feature as patentable, novel, non-obvious, or legally protectable without a separate prior-art and legal review. Preserve invention chronology, technical distinctions, experiment records, and public-disclosure dates for later assessment. ## Cold-Start Test Instruction For controlled tests, give the receiving AI the unchanged deployment package and only a bounded task instruction. Do not separately explain the deployment topology or governance rules that this context file is intended to reconstruct. Compare the resulting package against a predeclared QA rubric. ## Required Post-Change Checks - Correct version advancement. - Correct custom-domain target. - Sitewide version footer. - `/deployments/` page and `/deployments/feed.xml`. - Machine registries parse as JSON. - RSS/XML parses. - Internal links resolve. - Mobile layout remains usable. - No `/functions/` directory is introduced into static production unless explicitly authorized after beta certification. - No secrets or credentials are embedded. - New controlled evidence remains clearly distinguished from certified claims. ## Handoff Principle **Institutional continuity should be carried by governed artifacts, not by dependence on a particular AI model, account, or conversation history.** ## Update — v2.7 Candidate 01 (October 18, 2026) - **Source of record:** `https://github.com/jermainewhitesideedu/jermainewhiteside-cv`. `main` starts with the unchanged live v2.5 Candidate 16 package. Deploy through Cloudflare Workers Builds to Worker `jermaine-whiteside-cv` (static assets; see `wrangler.jsonc` and `.assetsignore`). - **Rollback baseline:** v2.5 Candidate 16. - **New surfaces:** `/fellowships/`, `/references/` and `/cv/` (ported from v2.6 Test Candidate 20; v2.6 was never deployed, so numbering moves to v2.7). - **Receiver JW-CV-RCV-001:** `/evidence/external/`. `automation/receive_registers.py` fetches sender registers, validates them, stages entries as candidate evidence and appends to `/evidence/external/consumption-log.json`. It never approves. Owner decisions live in `/evidence/external/approvals.json`. - **Rule carried forward:** a received register entry is candidate evidence, not a certified CV claim. ## Update — v2.7 Candidate 02 (October 18, 2026) - **Deployment target (correction):** the repository deploys through the Cloudflare **Pages** project `jw-certified-cv` (production branch `main`), not Worker Builds. The earlier "beta" description of `jw-certified-cv` is superseded. - Pages publishes the whole repository, so `automation/`, `tools/` and `.github/` are public. They hold no secrets. Never commit credentials to this repository. - **Custom-domain check:** on October 18, 2026, `cv.jermainewhiteside.com` still served v2.5 Candidate 16 from the earlier target while `jw-certified-cv.pages.dev` served v2.7. A successful build is not proof of production promotion: always check the footer on the custom domain. - **Rollback baseline:** v2.7 Candidate 01. ## Update — v2.7 Candidate 03 (October 18, 2026) - Ships work committed to the Candidate 02 branch after its pull request was merged. Candidate 02's history entry is left as deployed. - New surfaces: global footer on every page; AI Failure Register at `/ai-failures/` (JSON + RSS, six-criteria gate). - **Process rule:** never push to a release branch after its pull request is merged. Start a new branch and a new candidate. - **Rollback baseline:** v2.7 Candidate 02. ## Update — v2.7 Candidate 04 (October 18, 2026) - Records PRs #4–#19, which changed production after Candidate 03 without release controls (see AI Failure Register entry 012). - Restores the homepage truncated by commit b00f78c (entry 011). - **Release rule:** every merge that changes served files must advance the footer version, add a deployment-history entry and release notes, and write a SHA-256 manifest. - **Rollback baseline:** v2.7 Candidate 03 as recorded; the Candidate 04 manifest is the first complete record of the current files. ## Update — v2.7 Candidate 05 (October 18, 2026) - One menu on every page, generated by `tools/unified_nav.py`: Home, Research, CV, Credentials, Evidence. Change menus or page labels there and re-run; do not hand-edit menus on individual pages. - One label per page (Printable CV, Evidence Archive, Fellowships, Certifications). - Machine-readable index at `/machine-readable/` (built by `tools/build_machine_index.py`). - **Rollback baseline:** v2.7 Candidate 04. ## Update — v2.7 Candidate 06 (October 19, 2026) - New planning page `/biosketch/nih-common-form/`, generated from `biosketch/nih-common-form/plan.json` by `tools/build_nih_plan.py`. After editing plan.json, re-run that tool, then `tools/unified_nav.py` and `tools/global_footer.py`. - The page prepares content only. NIH accepts only the certified PDF produced in SciENcv; do not present this page as the biosketch. - **Rollback baseline:** v2.7 Candidate 05. ## Update — v2.7 Candidate 07 (October 19, 2026) - Adds CITI record 70645215 to the CITI register (16 records). Certificates are public; completion reports stay controlled. - **Rollback baseline:** v2.7 Candidate 06. ## Update — v2.7 Candidate 08 (October 19, 2026) - Adds CITI record 70645219 to the CITI register (17 records). Certificates are public; completion reports stay controlled. - **Rollback baseline:** v2.7 Candidate 07. ## Update — v2.7 Candidate 09 (October 19, 2026) - **CITI claim rule:** a CITI record claims only "Training completed". Do not call it a certification, licence, credential, appointment or proof of competence, and do not say it satisfies or meets a requirement. Name the evidence held and which document the verification link points to. `tools/citi_claims_c09.py` renders the cards from the registry. - **Rollback baseline:** v2.7 Candidate 08. ## Update — v2.7 Candidate 10 (October 19, 2026) - New page `/biosketch/coi-readiness/` from `biosketch/coi-readiness/readiness.json` (`tools/build_coi_readiness.py`). Publish only public roles, rules and template language; effort, compensation, ownership and contract amounts stay in the owner's private Drive worksheet. - **Rollback baseline:** v2.7 Candidate 09. ## Update — v2.7 Candidate 11 (October 19, 2026) - Ownership (EWTECHNERD LLC 100%, Anointed Holdings LLC 100%, as of 2026-10-19) lives once in `biosketch/coi-readiness/readiness.json` under `ownership`; update it there when it changes and rebuild. - Compliance Readiness Score: computed by `tools/build_coi_readiness.py` from `compliance_score.requirements` (met 1, in progress 0.5, not met 0). Change a status and rebuild; the score follows. - **Rollback baseline:** v2.7 Candidate 10. ## Update — v2.7 Candidate 12 (October 19, 2026) - Grant conflict register now lists five organizations: Anointed Connect Church Inc., EWTECHNERD LLC, Anointed Holdings LLC, IAM of America and Reinvest Carolina Inc. Roles for the two nonprofits come from the owner's statement (IAM of America) and reinvestcarolina.com (Reinvest Carolina Inc.). - **Rollback baseline:** v2.7 Candidate 11. ## Update — v2.7 Candidate 13 (October 19, 2026) - The printable CV (`cv/dynamic-cv.js`) reads CITI training from `/evidence-cv/citi-research-training/registry.json`. Add CITI courses to that registry only (via the citi tools); do not add them to `cv/data/base-cv.json`. - CITI register: 18 records. - **Rollback baseline:** v2.7 Candidate 12. ## Update — v2.7 Candidate 14 (October 19, 2026) - New page `/biosketch/sciencv/` (tool `tools/build_sciencv_print.py`, data `biosketch/sciencv/common-form.json`): SciENcv-layout Common Form print preview. Fields not on record stay `[to confirm]`. - **Rollback baseline:** v2.7 Candidate 13. ## Update — v2.7 Candidate 15 (October 19, 2026) - `wrangler.jsonc` is the source of truth for plain-text variables and the `RESEARCH_CACHE` KV binding (production and preview namespaces). Secrets live only in the Cloudflare dashboard. - `GET /api/research/status` reports setting presence and credential validity as states; it never returns a value. - **Rollback baseline:** v2.7 Candidate 14. ## Update — v2.7 Candidate 16 (October 19, 2026) - `GET /api/research/profile` (functions/api/research/profile.js): ORCID public works and PubMed records, source-reported, cached in RESEARCH_CACHE. PubMed matching uses the ORCID [auid] search or `/research/connected/pubmed-allowlist.json` only. - Display: Connected Research Profiles on `/orcid/`. Imported records never enter certified JSON-LD, evidence RSS or the certified CV. - Imports on in Preview, off in Production. - **Rollback baseline:** v2.7 Candidate 15. ## Update — v2.7 Candidate 17 (October 19, 2026) - `/api/research/profile` rebuilds its snapshot when settings change and retries failed sources after 30 minutes. - **Rollback baseline:** v2.7 Candidate 16. ## Update — v2.7 Candidate 18 (October 19, 2026) - ORCID and PubMed imports on in Production. Records on `/orcid/#connected-research` are source-reported only. - **Rollback baseline:** v2.7 Candidate 17. ## Update — v2.7 Candidate 19 (October 19, 2026) - Imported titles are decoded and trimmed; snapshot cache key is `snapshot:v2`. - **Rollback baseline:** v2.7 Candidate 18. ## Update — v2.7 Candidate 20 (October 19, 2026) - Homepage and Publications link to `/orcid/#connected-research` with the live ORCID count (`research/connected/orcid-summary.js`). - ORCID page record check uses `/research/connected/site-publications.json`; re-run `tools/build_site_publications.py` after editing the Publications page. - PubMed line hidden until it has matches. - **Rollback baseline:** v2.7 Candidate 19. ## Update — v2.7 Candidate 21 (October 19, 2026) - USC doctoral study: May 2023 – Fall 2023, outcome Transfer (to Liberty University). - **Rollback baseline:** v2.7 Candidate 20. ## Update — v2.7 Candidate 22 (October 19, 2026) - CITI register: 19 records (70645238 Partnering with Technology Companies added). - **Rollback baseline:** v2.7 Candidate 21. ## Update — v2.7 Candidate 23 (October 19, 2026) - Adds /research/maps/: five connected strands, 15 proposed directions, linked publication foundations, and a staged scholarly journey. - Map data: research/maps/registry.json; regenerate with tools/build_research_maps.py, then unified_nav.py and global_footer.py. - Research plans remain distinct from certified CV claims, study approvals, and journal acceptance. - **Rollback baseline:** v2.7 Candidate 22. ## Update — v2.7 Candidate 24 (October 19, 2026) - Adds /research/maps/#methodology with a proposed research approach and an expandable Accounting Lab protocol summary. - Public method content is generated from research/maps/registry.json by tools/build_research_maps.py. - Protocol remains prospective; no study approval, evaluation result, journal acceptance, or validated instrument is asserted. - **Rollback baseline:** v2.7 Candidate 23. ## v2.7 Candidate 25 — synchronization - Runs EWTECHNERD and E-Vision candidate intake. - Adds checked ORCID runtime status, governed SSRN catalog synchronization, Praxis export intake and capability-feed validation. - Reports failures and stale state without altering approved CV claims. - Whiteside Research stays blocked until its canonical source endpoint is supplied. ## v2.7 Candidate 26 — synchronization - Runs EWTECHNERD and E-Vision candidate intake. - Adds checked ORCID runtime status, governed SSRN catalog synchronization, Praxis export intake and capability-feed validation. - Reports failures and stale state without altering approved CV claims. - Whiteside Research stays blocked until its canonical source endpoint is supplied. ## v2.7 Candidate 27 — synchronization - Runs EWTECHNERD and E-Vision candidate intake. - Adds checked ORCID runtime status, governed SSRN catalog synchronization, Praxis export intake and capability-feed validation. - Reports failures and stale state without altering approved CV claims. - Whiteside Research stays blocked until its canonical source endpoint is supplied. ## v2.7 Candidate 28 — synchronization - Runs EWTECHNERD and E-Vision candidate intake. - Adds checked ORCID runtime status, governed SSRN catalog synchronization, Praxis export intake and capability-feed validation. - Reports failures and stale state without altering approved CV claims. - Whiteside Research stays blocked until its canonical source endpoint is supplied. ## v2.7 Candidate 29 — synchronization - Runs EWTECHNERD and E-Vision candidate intake. - Adds checked ORCID runtime status, governed SSRN catalog synchronization, Praxis export intake and capability-feed validation. - Reports failures and stale state without altering approved CV claims. - Connects the confirmed Whiteside Research WIPS record index as contextual candidate evidence. ## v2.7 Candidate 30 — synchronization - Runs EWTECHNERD and E-Vision candidate intake. - Adds checked ORCID runtime status, governed SSRN catalog synchronization, Praxis export intake and capability-feed validation. - Reports failures and stale state without altering approved CV claims. - Connects the confirmed Whiteside Research WIPS record index as contextual candidate evidence. ## v2.7 Candidate 31 — Owner decisions on received CV evidence - Records owner approval of 24 EWTECHNERD entries as qualified, founder-attributed portfolio evidence. - Places three E-Vision candidates on hold for role, entity, speaker or date verification. - Preserves 18 context-only records and 12 sender exclusions in the reviewed batch. - Approvals are bound to source content hashes; no automatic rewrite of certified or printable CV claims. ## v2.7 Candidate 32 — Owner decisions on received CV evidence - Records owner approval of four Whiteside Research updates as authored public scholarship and one narrated presentation. - Retains two authority records and one research-program record as context only. - Preserves article publication/page-creation dates, non-peer-reviewed classification, AI disclosure and bounded pilot interpretation. - Preserves earlier 24 EWTECHNERD approvals, three E-Vision holds, and sender exclusions; no automatic certified or printable CV rewrite. ## v2.7 Candidate 33 — CV source synchronization - Connects the Lost Estate Research Laboratory public-record bundle to JW-CV-RCV-001 with role candidates and context-only case, evidence and capability records. - Preserves source-reported status, registration-pending evidence and limits on grant or university claims. - Runs EWTECHNERD and E-Vision candidate intake. - Adds checked ORCID runtime status, governed SSRN catalog synchronization, Praxis export intake and capability-feed validation. - Reports failures and stale state without altering approved CV claims. - Connects the confirmed Whiteside Research WIPS record index as contextual candidate evidence. ## v2.7 Candidate 34 — CV source synchronization - Connects AnointedConnect.com and acclab.anointededu.com as separate CV evidence sources. - Preserves the church preview and its declared .org canonical origin as context, without changing host settings or approvals. - Stages dated lab responsibilities for review and retains capability and bylaw records as context; no implementation or adoption inferred. - Preserves all prior owner decisions and the weekly intake workflow. ## v2.7 Candidate 35 — Owner decisions on received CV evidence - Records four hash-bound owner approvals for dated Anointed Connect Church laboratory responsibilities. - Groups governance leadership development, evidence repository work, documentation review and dashboard planning beneath the existing Senior Pastor role in the printable CV. - Preserves September 27, 2026 assignment dates and source limits; no task completion, measured outcome or bylaws adoption inferred. - Retains church preview and remaining lab records as context, preserving all earlier decisions. ## v2.7 Candidate 36 — CV source synchronization - Adds the bounded ACC-IMP-008 Knowledge, Experience and Evidence Map reference contract and JSON Schema. - Reconciles source hashes, scoped owner approval and actual CV placement; rejects invented labels, references and approved uses. - Separates receiver hash prefixes from full normalized-entry and approval-decision digests; no raw-file immutability asserted. - Includes 20 contract and reconciliation tests; visualization, accessibility verification and empirical evaluation remain pending. ## v2.7 Candidate 37 — CV source synchronization - Adds the interactive Knowledge, Experience and Evidence Map for the reconciled ACC-IMP-008 reference. - Generates a selectable SVG and matching searchable relationship table from the same authority-checked data. - Exposes source dates, scoped owner decisions, source links, normalized hashes and interpretation limits; works as a readable table without JavaScript. - Adds three rendering checks to the existing 20 contract tests and integrates navigation and sitemap generation; no new CV claims or theory relationships approved. ## v2.7 Candidate 38 — CV source synchronization - Adds an independent researcher map of four approved self-published updates and one narrated presentation, with 16 nodes and 15 scoped relationships. - Reconciles each authored-output relationship against the actual WRS record, owner decision, author, canonical URL and classification; no new approvals or theory links inferred. - Fixes phone display with stacked labeled relationship cards, full-width controls, wrapping evidence hashes and a selectable mobile record list. - Checks diagram/table equivalence, approval boundaries and mobile field labels; browser checks at 375px and 415px content widths found no horizontal page overflow, including expanded evidence details. ## v2.7 Candidate 39 — CV source synchronization - Restores the omitted REA 2026 paper using the official program title and July 7 session date. - Adds Conference Presentations to the dynamic CV, sourced from explicitly approved academic-activity records. - Synchronizes the academic evidence page, search index, source registry and CV correction ledger; preserves source evidence and owner-reported delivery as distinct. - Does not conflate the conference paper with a later differently titled podcast or assert peer-reviewed journal status. ## v2.7 Candidate 40 — CV source synchronization - Adds Corporate Infrastructure and Applied Technology as the third evidence map, using 23 approved qualified portfolio projects and the approved founder statement. - Reconciles 70 nodes and 69 relationships against source entry hashes and scoped owner decisions, retaining AI assistance, component authorship and historical deployment limits. - Provides project selection with a focused evidence path, a searchable full-portfolio relationship list, and the existing mobile record chooser and labeled cards. - Adds seven corporate reconciliation tests and a rendering equivalence check; no new CV claims, owner approvals, current deployment certifications or theory links introduced. ## v2.7 Candidate 41 — CV source synchronization - Corrects corporate project diagram switching by toggling SVG hidden attributes explicitly. - Keeps the selected diagram, node details and relationship list synchronized across project selection and full-list mode. - Preserves all portfolio approvals, evidence limits, mobile cards and existing research and pastoral views. ## v2.7 Candidate 42 — CV source synchronization - Versions corporate map assets so browsers fetch the corrected SVG-switching script. - Preserves the existing 23 project records, 69 relationships and all approval and interpretation limits. ## v2.7 Candidate 43 — CV source synchronization - Restores Golden Key honor-society membership to the dynamic CV with direct credential verification, certificate and badge links. - Records a live verification observation and preserves the distinction between a membership credential and a separate award. - Adds the accepted AERA Graduate Student Council newsletter contribution using public bibliographic metadata only; final publication remains unverified. - Keeps the private acceptance screenshot and review-copy URL out of public files; synchronizes academic evidence, search metadata and the CV correction ledger. ## v2.7 Candidate 44 — CV source synchronization - Corrects the newsletter collection variable so the dynamic CV renders Golden Key honors and accepted contributions. - Adds a runtime synthesis check using actual repository data and versions the CV script to refresh cached clients. ## v2.7 Candidate 45 — CV source synchronization - Adds the Research Compliance & Institutional Board map with hash-bound training records and explicitly qualified IRB review status. - Updates homepage Knowledge Map and Evidence Register buttons, Golden Key membership badge, SSRN snapshot and AERA completed-review count. - Retains Golden Key membership and accepted AERA newsletter contribution on the dynamic CV; no final publication or extra award inferred. - Updates owner-reported IRB Under Pre-Review status without implying approval, and stacks knowledge-map navigation for iPhone widths. ## v2.7 Candidate 46 — CV source synchronization - Corrects IRB status to Liberty University's October 19 No Human Subjects Research determination, including application safeguards and modification limits. - Records the stale-status correction in the AI Failure Register, with owner-reported ChatGPT/Codex model 6.1 attribution and the six-criteria RSS gate preserved. - Adds an ecosystem status page, honest endpoint checks, incident feed and expanded trust/legal navigation. ## v2.7 Candidate 47 — Reviewed continuing education on the dynamic CV - Adds four owner-reviewed continuing-education records to the dynamic CV, including Harvard Medical School participation, IsraelX/edX course completion and two NonprofitReady certificate programs. - Reuses existing credential identifiers, publishes the two NonprofitReady supporting certificates, and preserves exact completion dates and qualification limits. - Clarifies that Harvard activity-designated credits do not establish individually earned credits.