GracEleoJohn.com · Portfolio · 2026
AI Analytics & Agentic Workflow Designer · MBA Candidate
Write the vision and make it plain.Habakkuk 2:2
Eleven-plus years at the US Mission in Abuja, Nigeria, coordinating travel across a large operation, taught me what most analytics training does not: a process is only as good as the people who have to live inside it every day. Everything since has been built on that foundation.
Three degrees have sharpened the same lens through different disciplines — an AS in Hospitality and Tourism Management, a BAS in Supervision and Management with a Healthcare Administration major, and now an MBA in Business Analytics at Palm Beach Atlantic University. Operations taught me how work actually flows. Healthcare taught me what happens when it does not. Analytics gave me the means to measure both.
I am building toward AI operations and agent design — designing agent workflows, testing where they hold up, and documenting what the results actually show.
Palm Beach Atlantic University
Major in Healthcare Administration
Foundation in operations, service, and global coordination
Process improvement and operational analytics · Coursera
Coursera and Kaggle · Ongoing development in agent design and applied analytics
Institute for Operations Research and the Management Sciences (INFORMS) · International Institute of Business Analysis (IIBA)
Applied work from coursework, institutional analytics, and independent build projects.
A twelve-item diagnostic that scores organizational AI readiness across six dimensions and identifies the binding constraint — the single weakest capability that would block adoption first. Runs entirely in the browser; nothing is collected.
Open the instrument →Designed and maintained a reporting framework for PBAU's Warren Library — staff efficiency, patron volume and timing, study room utilization, and floor activity — delivered directly to library leadership. Included a holdings reconciliation merging roughly 139,000 records.
A journaling and goal-tracking application rooted in Habakkuk 2:2, built as the applied project for my Business Intelligence coursework under overseeing professor Dr. Ernesto Lee. It is the sandbox where I design and test the agent pipeline below. Free to use.
Oral presentation — "Writing the Vision in a Digital World: Renewing Christian Stewardship Through Data-Driven Analytics." 15th Annual PBAU Interdisciplinary Research Conference, alongside mentor Dr. Ernesto Lee.
Data science and machine learning projects against real-world datasets, added as completed. Focused on the analytical fundamentals underneath agent design rather than tooling alone.
This site — a self-contained HTML, CSS, and JavaScript portfolio designed, architected, and deployed independently.
The design insight that reshaped this work: individual agents are easy and largely useless. The value is in connecting them into a single pipeline where each hands off to the next and the last one feeds back to the first. WriteTheVZN is the sandbox where that pipeline runs.
City on WhatsApp and City Audio handle inbound interaction, routing, and fallback. Both live.
Content and Repurposing agents turn one devotional into five formats; the Video Agent renders the short-form output.
Analytics and ASO agents close the loop — performance data and store-listing signals feed the next content cycle.
WhatsApp conversational agent — handles inbound questions, guidance, and routing.
Voice companion in the app — spoken interaction, prompts, and response.
Drafts captions, carousels, devotionals, and newsletters from a shared brief. Output formatting under refinement.
Expands a single devotional entry into carousel, short script, caption, email blurb, and blog post.
Renders short-form video from written content, including scripture sets for children.
Weekly plain-language performance summaries from GA4 and Metricool — the feedback loop into content.
App Store and Play listing optimization — keyword tracking, screenshot copy testing, review monitoring and drafted responses.
Comment and message handling. Deliberately deferred — it has nothing to respond to until the publishing pipeline is running.
Voice modeling study — brand voice consistency across long-form output.
Deciding what not to build is the harder half of this discipline. The Engagement Agent is deferred because a responder with nothing to respond to is wasted effort, and sequencing like that is what separates a working system from a collection of demos.
I am seeking an internship in AI operations, business analytics, or process design, and I am open to research collaboration and speaking on applied AI in operations.
contact@graceleojohn.com