Techno-functional leader. Long career leading product development in healthcare IT, now working full-time and independently on FHIR (payer-side), openEHR, MCP, agentic LLM workflows, and Radiology AI — building open-source reference implementations.
I work in AI-native, plan-mode-first workflows, and I read, write, and review code throughout the build (now AI-augmented).

_{🩺 Healthcare IT — long career  ·  👥 Led teams of 30+ engineers across multiple geographies  ·  🔌 FHIR (R4/R5, Da Vinci) · openEHR · MCP  ·  🤖 LangGraph · LangChain · MONAI · MedGemma  ·  🧠 AI-augmented builder (Copilot · Claude Code)}

I work at the intersection of healthcare standards and applied AI — designing systems, writing the specifications, and getting them into a working state end-to-end:

• 🧭 Solutions Architecture for Clinical AI — system design and reference implementations spanning data, agents, evals, and clinical-workflow fit
• 🔌 Healthcare Interoperability — FHIR R4/R5 (with emphasis on the payer side — Da Vinci PAS, CRD, DTR, PDex, Plan-Net, Drug Formulary, BCDA), openEHR (CKM, EHRbase, AQL), SMART-on-FHIR, HL7
• 🤖 Agentic Clinical AI — multi-agent workflows on LangGraph / LangChain, MCP servers and clients for clinical data and policy reasoning, RAG over clinical and payer-policy corpora
• 🩻 Radiology AI — early-stage detection of lung disease and breast cancer; DICOM, MONAI, MedGemma, Orthanc PACS, OHIF (current independent engagement, under NDA)
• 👥 Techno-functional leadership — long career leading product development on Sunrise Clinical Manager, Sunrise Surgery, and adjacent products at Altera Digital Health (formerly Allscripts) with teams of 30+ engineers across multiple geographies
• 🧠 AI-native, plan-mode-first workflows — daily driver of GitHub Copilot and Claude Code for plan-mode thinking, spec-driven builds, architecture review, and code review. I read, write, and review code throughout the build (now AI-augmented).

For most of my career I led product development on Sunrise Clinical Manager, Sunrise Surgery, and adjacent products at Altera Digital Health (formerly Allscripts) — teams of 30+ engineers across multiple geographies, shipping into live clinical environments.

• Delivery philosophy — spec-first and AI-orchestrated, the same way the repos on this profile are built. Plan-mode before code, ADRs for non-trivial decisions, eval and review gates per slice.
• Code-review and mentoring — I stay close to the code through reviews and architecture conversations, and back engineers when their judgment is sound. The goal is teams that can decide without me in the room.
• Hiring — I optimise for judgment, domain curiosity, and figure-it-out ability over framework counts. Same bar I hold myself to.

Since March 2025, I have been working full-time and independently, including a current fractional engineering-leadership engagement on a Radiology AI venture focused on early-stage detection of lung disease and breast cancer (under NDA).

Open to Healthcare AI Solutions Architect, Technical Leader, or Fractional / Advisory roles where deep healthcare-IT domain expertise meets applied AI.

Sorted by relevance, not date. See pinned repos below or my full repository list.

Browse all repos by topic: #fhir · #agentic-ai · #mcp · #langgraph · #healthcare · #openehr · #clinical-ai

Languages

Backend & Cloud

Healthcare Standards

AI / ML

AI-Augmented Workflow (daily drivers)

My working assumption is that plan-mode, spec-driven, AI-orchestrated workflows are now the senior norm — not a quirk. Every repo on this profile is built this way, and this is how I expect the teams I lead to ship.

1. Plan-mode first — talk through the problem, constraints, and trade-offs with Claude Code or Copilot agent before writing a line of code. The plan is the artefact.
2. Specification-driven — design doc, sequence diagram, FHIR resource map, agent graph, or eval plan produced with the AI, then reviewed critically against domain context.
3. Build in small slices — each slice reviewed for correctness, security (OWASP), and clinical safety. AI as reviewer; judgment stays with me.
4. Evals and documentation as first-class outputs — every repo ships with a real README, measurable behaviour, and a clear status (WIP / Stable / Reference).

As a leader, my job is to set up the quality gates a team ships against — the spec rituals, eval bars, ADR cadence, and code-review standards — not to be the fastest typist in the room.

For hiring conversations: I'm strongest in architecture rounds, system-design discussions, and walking through any of the repos on this profile. If your loop is built around live algorithm whiteboarding, we're probably not the right fit — and that's a useful filter for both of us.

• 🧭 Prior-Auth Co-pilot (flagship #1, in build) — agentic, FHIR-native PA co-pilot for the CMS-0057 Jan 2027 mandate. Da Vinci PAS / CRD / DTR + policy reasoning + audit trail. Public roadmap and weekly slices in progress.
• 🧪 Clinical LLM Quality Harness (flagship #2, in build) — eval & observability framework across three tracks: ambient-scribe note quality, prior-auth reasoning quality, and clinical Q&A grounding.
• 🩻 Radiology AI (NDA, ongoing) — fractional engineering leadership on early-stage detection of lung disease and breast cancer; DICOM, MONAI, MedGemma, Orthanc PACS, OHIF.
• 🏗️ fhir-dqm-engine (Pramana) — care-gap API + AI layer on top of the CQL quality-measure engine; 69.8% BP control rate measured on a 279-patient synthetic cohort.
• 🛠️ fhir-mcp-suite — extending MCP server coverage for more FHIR resources; feeds the Prior-Auth flagship.

• 💼 LinkedIn: https://linkedin.com/in/paragmedsinge
• 📧 Email: paragmedsinge@yahoo.com
• 🌍 Based in: Pune, Maharashtra, India · open to remote / hybrid worldwide
• 💬 Open to Healthcare AI Solutions Architect, Technical Leader, or Fractional / Advisory roles where deep healthcare-IT domain expertise meets applied AI.

_{⚡ Note: The repos on this profile are reference implementations and working prototypes built around real interoperability and clinical-AI problems — not tutorials. Each is clearly labelled WIP / Stable / Reference.}