Physician-Informaticist deploying HIPAA-compliant clinical AI at scale.
I architect and ship production AI on Anthropic Claude (AWS Bedrock) for 40+ post-acute facilities, pairing five years of frontline medicine with health informatics and hands-on engineering.
Essen Healthcare · Bronx, NY
A clinician who builds.
I spent over five years as a practicing medical doctor, managing 50+ daily patient encounters, before moving into health informatics in the US, earning an MS at Sacred Heart (GPA 3.92) and an AI in Healthcare credential from Johns Hopkins.
That clinical grounding is the point. I don't build healthcare AI from the outside looking in. I build it knowing exactly where documentation breaks, where clinicians lose time, and where a tool has to earn trust before it earns adoption.
Today, as Clinical Informatics Analyst - AI & Automation at Essen Healthcare, I architect production clinical AI serving ~10,000 SNF patients, lead a team of 17, and train ~50 clinical staff on the tools I build. My focus is clinical AI that earns trust: documentation that's accurate, compliant, and actually gives time back at the point of care. I build platform-agnostic by default, with production fluency across 6+ EHR systems, so the work travels with the environment rather than the vendor.
My path
Where I've worked.
With Essen since Feb 2025, promoted into this role after building the organization's clinical AI capability from the ground up.
- Architected a HIPAA-compliant clinical documentation platform serving ~10,000 SNF patients on Anthropic Claude (AWS Bedrock), auto-generating E&M-compliant notes across APCM, AWV, monthly, and specialty visits with embedded CPT/HCPCS suggestions.
- Hardened the platform to pass corporate security review with field-level PHI encryption, JWT/refresh-token auth, and role-based access across 4 provider roles, resolving 90 audit findings.
- Built an AI HCC risk-adjustment gap-analysis tool on CMS-HCC V28 (8,019-code ICD-10 crosswalk) surfacing undercoded diagnoses with verbatim evidence and per-patient RAF gain across 6+ EHR systems.
- Lead a functional team of 17 Clinical Assistants and deliver AI adoption training to ~50 staff across 40 facilities.
- Optimized clinical workflows across SigmaCare, PointClickCare, WellSky, eClinicalWorks, and Allscripts, improving data accuracy, interoperability, and care coordination across post-acute facilities.
- Strengthened documentation and intake processes to support high-quality downstream clinical decision-making and value-based care initiatives.
- Managed 50+ daily patient encounters; built structured documentation frameworks that improved patient satisfaction by 25% and reduced follow-up inquiries.
- Led COVID-19 protocol training to 100% staff completion and launched a patient resource guide that cut appointment no-shows by 15%.
- Managed and documented a broad range of complex cases across multiple clinical departments.
Systems I've built and shipped.
Production clinical AI designed, built, and deployed across the post-acute network.
Clinical Documentation Platform
HIPAA-compliant platform serving ~10,000 SNF patients that auto-generates E&M-compliant notes on Anthropic Claude (AWS Bedrock) with embedded CPT/HCPCS code suggestions, hardened to pass corporate security review.
HCC Risk-Adjustment Gap Analysis
An AI tool on the CMS-HCC V28 model (8,019-code ICD-10 crosswalk) that surfaces undercoded diagnoses with verbatim source evidence and per-patient RAF gain, with guardrails preventing code fabrication across 6+ EHRs.
Clinical Note Agents
A suite of Copilot Studio agents drafting structured notes for Annual Wellness Visits, APCM/GPCM care management, and pulmonology consults, refined iteratively against real clinician feedback.
Operational Dashboards
Real-time dashboards giving nursing-home division leadership visibility into provider productivity, encounter volume, and facility census trends across the network.
The Second Read.
Built end to end in a five-hour window at the Abridge x Anthropic x Lightspeed Healthcare AI Hackathon in San Francisco, July 2026. A working verification agent for the documentation problem I work on every day.
Reads the note back against the chart before it is signed.
Ambient scribes treat the doctor and patient conversation as ground truth. The Second Read is the check on the other side. It reconciles a physician's draft note against PT, nursing, OT, and consult notes, flags any statement the rest of the chart contradicts, and drafts corrections the physician approves one flag at a time. Built for post-acute care, where functional mobility documentation drives PDPM reimbursement and discharge safety.
Gate at sign
Research that shapes how I build.
Capstone research from my master's program, recognized by New England HIMSS, distilled into a short, readable brief while the full manuscript is prepared for journal submission.
AI in Breast Cancer Detection and Diagnosis: A Systematic Review
What 31 peer-reviewed studies from 2020 to 2024 show about AI's real impact on the accuracy and efficiency of breast cancer detection.
Read the research brief →“AI can play a transformative role in breast cancer diagnostics, but it is most effective when used as a decision-support tool rather than as a replacement for human expertise.”
Credentials.
Let's build something that helps clinicians.
Open to conversations about clinical informatics and healthcare AI. I'd be glad to connect.