AI Strategy & Implementation
Leading enterprise AI adoption for a health system — balancing capability, cost, and data governance across on-prem and cloud AI infrastructure.
AI in healthcare needs an operator, not just a vendor slide
Enterprise AI in a clinical setting isn't just a model choice — it's an infrastructure, governance, and change-management problem. The same discipline I bring to Epic and cloud infrastructure now extends to how our health system deploys AI safely and cost-effectively.
On-Prem Inference Capacity
Evaluating and standing up on-premises inference infrastructure for workloads where data residency, latency, or cost make cloud-only AI the wrong answer — sized and governed like any other critical system.
AWS Bedrock & Managed AI
Leveraging AWS Bedrock for managed foundation model access where cloud deployment fits — keeping model selection, cost, and security posture aligned with enterprise governance standards.
RAG & Knowledge Bases
Designing retrieval-augmented generation pipelines and Knowledge Bases that ground AI responses in an organization's own policies, clinical documentation, and operational data — not just a model's training data.
Agentic Workflows
Building agents that can take multi-step action across enterprise systems — with the guardrails, logging, and human-in-the-loop checkpoints a healthcare environment requires.
AI Governance & Risk
Extending the same IT governance and cybersecurity discipline used for Epic and cloud infrastructure to AI: data handling, model access controls, and audit trails for every AI-assisted decision.
Team Upskilling
Built an incentive program to upskill infrastructure and application teams on cloud and AI technologies — treating AI fluency as a core competency for the whole department, not a specialist silo.
How I think about enterprise AI
Most organizations' first instinct with AI is to chase the model. Mine is to start with the infrastructure and governance question: where does inference need to run, what data is it allowed to touch, and how do we prove that after the fact.
That means treating on-prem inference capacity as a real capacity-planning exercise, not an afterthought — and treating AWS Bedrock, RAG pipelines, and Knowledge Bases as production systems with the same change-control rigor as any other clinical application.
It also means building agentic workflows with the same caution I'd apply to any automation touching patient data: clear scope, logging, and a human checkpoint wherever the stakes are high.
Core Building Blocks
- InferenceOn-Prem & Cloud
- PlatformAWS Bedrock
- RetrievalRAG & Knowledge Bases
- AutomationAgentic Workflows
- OversightGovernance & Audit
Standing up AI capability at your organization?
Let's talk about what an operationally sound AI rollout looks like for a healthcare environment.
Let's Talk