Selected Work
Six examples of technology built to produce a business result.
A business enablement platform: executive intelligence over every system a company runs.
Situation:
Mid-market companies run marketing, sales, finance and operations on separate systems that disagree on the numbers. Executives get stale reporting, attribution gaps between spend and closed revenue, and a data-team bottleneck between every question and its answer.
What I did:
Architected a system-agnostic intelligence layer that connects to CRM, ERP, ad platforms, call tracking, EDI and finance systems and organizes the combined data as a knowledge graph (taxonomy, ontology, graph). On top of it, Ask AI: a natural-language interface that answers cross-system questions with cited sources and a confidence rating, using retrieval over the knowledge graph and eight domain agents (finance, sales, marketing, operations, compliance and others) that carry each function's vocabulary and rules. Nightly-retrained forecasting for close probability and period-end outcomes, continuous monitoring with anomaly alerts, decision simulation against historical data, a learning loop that improves from plain-language corrections, and autonomous agents for follow-ups, digests and revenue attribution. Isolated per-customer databases, full audit trail, SOC 2 and GDPR ready, deployable to Azure, AWS, GCP or on-premises in two to four weeks.
Outcome:
Built in 2025 and now live with multiple customers. First cross-source insights within 30 days of deployment; customers report double-digit gains in marketing efficiency within a quarter and six-figure recoveries of closed-won revenue that had never been invoiced.
Multi-tenant B2B commerce platform, 200+ enterprise clients
Situation:
A B2B e-commerce platform with a single ERP integration and long client deployments, in a market where every prospect ran a different back office.
What I did:
Owned architecture and roadmap for eleven years. Re-platformed onto Azure as a proprietary multi-tenant system, diversified from one ERP integration to six, and rebuilt onboarding and the quote-to-order workflow between sales and engineering.
Outcome:
Client deployment time cut by 30 days. Outbound sales cycle accelerated 200%. Roughly $3.5M a year of rework removed. Platform run at 99.99% uptime with incident response under ten minutes. Company revenue doubled over the tenure.
SAP partner channel, $8M a year
Situation:
Growth depended on reaching mid-market manufacturers and distributors who bought through their ERP implementation partners, not directly.
What I did:
Built relationships and a technical partnership program with 20 of the top SAP implementation partners, including integration certification, joint solution design and partner enablement.
Outcome:
A channel producing $8M in additional annual revenue, and two invited talks at SAP partner conferences on ERP-integrated B2B commerce.
SOC 2 Type II and the security program behind it
Situation:
Enterprise clients and prospective buyers of the company needed evidence of security maturity, not assurances.
What I did:
Built the security program from the identity layer up: Microsoft Entra ID, Intune, Defender, conditional access, endpoint detection and response, and the policies and evidence collection to support an audit.
Outcome:
Passed the SOC 2 Type II audit cycle. Security posture became a selling point in enterprise deals and a clean line item in the company's sale diligence.
Enterprise AI framework for a healthcare organization
Situation:
A nonprofit health cost-sharing organization serving several hundred thousand members wanted to use AI across IT without creating governance or member-data risk.
What I did:
On a one-year fixed-term fractional CTO contract, introduced an enterprise AI framework covering use-case selection, data governance, model evaluation and vendor criteria, and led implementation in strategic areas of IT alongside business leadership.
Outcome:
A repeatable process for choosing and running AI initiatives in a regulated, member-data environment, delivered on scope and on schedule.
Clinical data program for an AI medical device (current)
Situation:
An early-stage company building an endoscopic ultrasound decision-support system for pancreatic cancer detection needed clinical data infrastructure that would survive regulatory, clinical and investor scrutiny.
What I did:
As fractional CTO, built the program from zero: IRB determination, a de-identified corpus with matched pathology, a physician-reviewed annotation protocol, leakage-proof dataset separation, and a competitive RFP to contract external ML development with IP separation terms. Present strategy and progress to investors.
Outcome:
Contributed to a Cleveland Clinic Innovations co-development proposal and two NIH R41 submissions.