AI Engineer, Fractional CTO — Workforce Intelligence Group, Entrepreneur, and Thought Leader building at the intersection of AI governance, MLOps, and enterprise innovation.
I enter complexity, create structure, align people and execution — and outcomes improve. That pattern has held across 26 years in tech and 17 years at IBM, from mainframe engineering to enterprise AI architecture.
I've designed end-to-end AI solutions for enterprise clients, rescued failing AI initiatives under pressure, and influenced multi-million dollar deals through technical credibility. I've owned outcomes for 70+ accounts across industries, specializing in Financial Services with banks generating $100M+ in yearly revenue. I built enablement programs for 300+ global attendees and mentored the next generation of engineers along the way.
My work has been recognized with IBM's Eminence & Excellence Award, 7 Manager's Choice Awards, IBM STAR Leadership selection, and the McKinsey Senior Management Accelerator. I don't just build systems — I build the people and operating models that make them last.
Leading technology strategy and innovation for AI-powered workforce solutions. Driving enterprise modernization and responsible AI adoption.
Designing end-to-end AI solution architectures for enterprise clients. Driving customer efficiency improvements, cost reductions, and measurable ROI. Rescued failing AI initiatives — improving model accuracy from 20% to 80%+.
Led strategic IT/AI roadmaps, infrastructure modernization, and cloud migrations for enterprise clients, with Financial Services banks generating revenue exceeding $100M yearly.
Managed 70+ SMB clients across industries, specializing in Financial Services, contributing to $2M ACV. Built 12 executive dashboards and drove a 25-35% reduction in unworked CTAs.
Led migrations, resolved 1,000+ defects in DB2 tools for z/OS, and delivered global technical enablement. Co-led bootcamps for 300+ global attendees with hands-on labs.
Built automation and testing frameworks for mainframe systems. Remediated 50+ field-reported issues, developed 29 automated scripts, and converted legacy assembler logic to Java.
Provided deep technical troubleshooting for complex enterprise systems and built the engineering discipline that everything else stands on.
AI models were at 20% accuracy weeks before go-live. Led rapid technical intervention — retrained models, improved training data quality, optimized configuration. Launch proceeded on schedule.
Contact center client needed to cut costs while improving customer experience. Designed the virtual agent end to end with measurable KPIs — 30% lower cost, 90% first-contact resolution, and more customers acquired.
Supported multi-million dollar enterprise deals including an $8.4M multi-year agreement, $2M ACV across 70 accounts, and multiple six-figure expansions — all driven by technical alignment and trust.
Co-led a global bootcamp and designed a virtual hands-on lab on AI orchestration and automation, treating enablement as a product delivery problem. Built standardized labs, RACI models, and live support infrastructure that became the repeatable framework for future programs.
When platform roadmap delays created churn risk, proposed alternate deployment paths using available capabilities. Customer retained, solution implemented, defection risk eliminated.
Led the modernization and migration of 300 applications, containerized on Red Hat OpenShift into production — the organization's first cloud-native deployment at scale, and the platform later workloads were built on.
Led the shift from static HR workflows to multi-agent AI — a supervisor orchestrator routing to LangGraph specialist agents over Workday, with guardrails, per-agent factsheets and an agent operations framework (orchestration patterns, guardrails, escalation) for workforce-intelligence use cases, from talent insights to employee self-service, governance designed in from day one.
View architecture →Led the move from ad-hoc notebooks to a governed, repeatable MLOps lifecycle for predictive models — data validation, experiment tracking, production-readiness gates, deployment, drift monitoring, and automated retraining on IBM Cloud Pak for Data.
View architecture →Deployed a conversational AI assistant for a healthcare organization during the pandemic — patient and staff questions answered at scale, real-world AI solving real-world problems under pressure.
Led the modernization and migration of 300 applications, containerized on Red Hat OpenShift into a production environment — the organization's first large-scale cloud-native deployment.
Migrated a contact center to an AI virtual agent: 75% containment, 88% intent recognition, and 90% first-contact resolution. Reduced costs 30% while acquiring more customers.
Designing compliance workflows, risk frameworks, and model monitoring systems. Making responsible AI the default, not the afterthought.
End-to-end model lifecycle management — from training to deployment to decommissioning — with full auditability.
Building phased training programs that take teams from dashboard basics to advanced exception handling and autonomous workflows.
Advising enterprise organizations on AI modernization, cloud migration, and operationalizing data science at scale.
Crafting technical demos that translate complex AI capabilities into clear business value for clients and stakeholders.
Publishing insights on AI augmentation, prompt engineering, and the evolving role of human expertise in automated systems.
Applied AI, developer enablement, practical tooling, and data science methodology. Arctic Code Vault Contributor. Cards marked ▶ include a short walkthrough of how the project works.
▶ Demo walkthrough
A web app to test Hugging Face LLMs — experiment with open-source models directly from the browser.
▶ Demo walkthrough
AI-driven insurance agent — intelligent automation for policy analysis, claims processing, and customer engagement. This build reads Attending Physician Statements for life-insurance underwriting: watsonx.ai (Mixtral-8x7B) summaries, hospitalization detection, risk classification, and RAG chat with the document.
▶ Demo walkthrough
Full-stack local AI workbench: intelligent document processing (PDF, Word, PowerPoint, text), video frame extraction and audio transcription, image OCR, vision analysis with LLaVA, and context-aware chat with local LLMs via Ollama — plus DOCX report export. Runs privately, on-device.
▶ Demo walkthrough
IBM Data Science capstone: segments neighborhoods across Atlanta, San Francisco, and New Orleans by venue mix (Foursquare API + scraped Wikipedia), picks k with the elbow method, maps clusters with Folium, and predicts campaign lift per customer persona.
▶ Demo walkthrough
Sample end-to-end MLOps implementation on IBM Cloud Pak for Data: seven sequential notebooks that take a model from data validation and feature engineering through experiment tracking, production-readiness checks, deployment to a deployment space, drift monitoring, and automated retraining.
Named inventor on patents and filings in enterprise AI and automation — ideas that moved from whiteboard to protected intellectual property.
[Problem Solved]
Outside of work I mentor and teach — from first lines of Python to the full data science methodology — with interactive tutorials anyone can run in the browser.
▶ Tutorial walkthrough
A graphical, interactive Python tutorial for absolute beginners. Learners launch it in the browser via Binder — no install — and work through variables, data types, lists, loops, conditionals, functions, and their first plot, one runnable cell at a time.
▶ Tutorial walkthrough
A graphical, interactive Python tutorial built around IBM's Foundational Methodology for Data Science. Learners work through all ten stages with a real automotive-industry use case, an algorithm-selection guide, and runnable cells.
Volunteer mentor with High Tech High Heels, supporting women and girls pursuing STEM — career navigation, technical growth plans, and the confidence to take the next step.
How smart teams leverage AI to amplify human expertise, not replace it.
Read on Medium →From human-guided development to self-healing systems.
Read on Medium →The surprising parallel between poor AI prompting and poor human communication.
Read on Medium →Information security strategies for organizations running OpenShift in regulated environments.
Read on Medium →Analyzing user feedback and escalation patterns to improve conversational AI performance.
Read on Medium →K-Means clustering and Folium data visualization for IBM Data Science capstone project.
Read on Medium →Resume, and the sample architectures behind the case studies.
Full professional background and credentials
PDF — DownloadReference designs I use as starting points with clients. Pick one to view it here, or download the full deck.
