CONVICTION / 01
Workflow systems, not features
Design the clinical workflow before the UI. Help that lives outside the workflow forces a context-switch, and clinicians under pressure won't take it.
Product Manager · AI Product Manager · Healthcare SaaS
I build AI products that meet clinicians inside the workflow — and turn healthcare complexity into systems that scale.
Five years shipping healthcare SaaS from 0→1 and operating it at enterprise scale, where the win is never the demo — it's adoption, compliance, and the edge cases underneath.
Based in Bengaluru, IN · open to PM / AI PM / Senior PO roles
Product Owner @ Evolent Health · formerly Senior PM @ 314e
Product Owner @ Evolent Health · formerly Senior PM @ 314e
80K+
Healthcare providers on a product I led
$45K+
Monthly recurring revenue, built 0→1
45%
Clinician productivity gain from GenAI
150K+
Monthly auth requests standardized via FHIR
01
How I build
Four convictions that show up in almost everything I ship — the reason my products get adopted instead of just launched.
CONVICTION / 02
Adoption over novelty
A brilliant feature nobody adopts is a failed feature. I optimize for time saved and clicks removed, not for the length of a release note.
CONVICTION / 03
AI as leverage, not autopilot
Use AI where it creates measurable leverage — search, answers, content creation — and treat accuracy, safety, and trust as first-class product responsibilities.
CONVICTION / 04
Outcomes over output
Prove value with analytics — ticket deflection, completion, revenue — not activity. Config over code, so the next customer onboards without another build.
Signature case study · 314e
Jeeves — just-in-time AI training, embedded inside Epic
A Just-in-Time AI training and support system for EHR users that lives inside the clinician's workflow. I led it from concept to launch in six months and scaled it to ~80,000 providers.
The reframe
The ask was “better training content.” The real problem was workflow distance.
Content can be excellent and still fail if a clinician has to leave the patient chart to find it. So I reframed the goal from producing more training to building an in-workflow enablement system: discoverable help at the moment of need, self-service that scales without adding headcount.
The bet I drove
Build Jeeves as an EHR-embedded system, not a standalone portal.
That meant going deep on Epic integration via SSO and SMART on FHIR, so clinicians reach help without a separate login or a workflow break. The trade was deliberate: I gave up a faster generic rollout across environments to win higher adoption through workflow-native access — the metric that actually mattered.
The system — six layers, one principle: bring help to the moment of need
Product interface preview unavailableAccessible inside Epic via SSO + SMART on FHIR — help arrives where clinicians already work.
Product interface preview unavailableNatural-language search surfaces the exact bite-sized video or tip sheet — no keyword guessing, no leaving the chart.
Product interface preview unavailableAn LLM bot grounded in the organization's content library answers instantly, then hands off to a live agent for complex cases.
Product interface preview unavailableAn AI toolbox — tip-sheet generation from video, screen recorder, AI narration, image editor — lets teams create more with less effort.
Product interface preview unavailableA review-comment-approve workflow with reminders and outdated-content flags — because trust is non-negotiable in clinical settings.
Product interface preview unavailableAssignments and in-workflow notifications turn EHR upgrades from a scramble into structured, tracked readiness.
The impact
$45K+
Monthly recurring revenue from the platform
~80K
Healthcare providers using Jeeves
~25%
Better onboarding via fewer support tickets and higher tutorial completion
45%
Clinician productivity gain — content creation from hours to minutes
Value was measured, not assumed: I stood up analytics on Matomo early, then migrated to Google Analytics as tracking maturity and scale demanded it.
Tradeoffs I owned
| Decision | Risk accepted | Why it was worth it |
|---|---|---|
| Deep Epic embedding (SMART on FHIR SSO) | Integration complexity | Adoption comes from workflow proximity |
| AI self-service via search + bot | Accuracy and trust expectations | Scale support without scaling headcount |
| Content automation (tip sheets, narration) | Change management for trainers | Cut creation cost, raise output speed |
| Governance workflow | Operational overhead | Prevent outdated, unsafe guidance |
“Jeeves made it easy for the Informatics team to fully own their content.”
Michelle F. — Manager, Clinical & Business Informatics · verified customer review
Case study 02 · Evolent Health
CarePro — configurable prior authorization at payer scale
As Product Owner, I own strategy and execution for enterprise prior-authorization capabilities serving large US healthcare payers — configurable platform logic, workflow automation, and API-enabled integrations with third-party payer systems.
PLAY01
Standardizing the un-standardizable reason code
The setup
Highmark had to transmit authorization reason codes outbound over FHIR, but internal codes mapped many-to-one onto the standardized set — with no clean way to decide which code wins, across 150,000+ requests a month.
The play
I designed a configurable reason-code ranking engine so payer-side precedence rules — not hardcoded logic — decide the winning code. Same engine, new payer, no rebuild.
The result
150K+ monthly requests standardizedclean outbound FHIR transmission
PLAY02
Fixing the SLA number everyone was contractually trusting
The setup
Contractual SLA reporting was being computed on flawed logic across 75 health plans — a live compliance and trust exposure hiding inside a third-party integration.
The play
I partnered directly with payer engineering to map system architecture and data hierarchies, traced the flaw to the SLA/TAT calculation, and redesigned the computation framework from the ground up.
The result
85% more accurate SLA reportingacross 75 health plans
PLAY03
Build once, onboard many
The setup
Every new enterprise payer meant another round of engineering. Integrations were being rebuilt per customer — an approach that couldn't scale with the pipeline.
The play
I built configurable, payer-level business rules — auth logic, SLA/TAT, workflow branching, integration behavior — as a reusable framework for scaling integrations instead of a per-customer rebuild.
The result
4 new enterprise payers onboardedwith zero added engineering
Also in the playbook
HIPAA-compliant MFA rollout
Led delivery of multi-factor authentication across 15 payer configurations, securing provider access while holding regulatory compliance.
80K+ users secured · 15 configurations
Pop & Stop automation
Designed workflow automation for Gold Card and PA-exempt scenarios, driven by pattern analysis of authorization outcomes.
48% fewer unnecessary authorizations
03
Impact, on the record
The numbers behind the work, in one place.
80K+
Healthcare providers on a product I led from 0→1
Scale
$45K+
Recurring monthly revenue generated by that product
Revenue
45%
Gain in clinician productivity from GenAI — hours to minutes
Adoption
85%
Improvement in contractual SLA reporting accuracy across 75 plans
Reliability
48%
Reduction in unnecessary authorization creation via automation
Efficiency
4
New enterprise payers onboarded with zero additional engineering
Leverage
04
The toolkit
What I reach for, grouped by the job it does.
AI & GenAI
- OpenAI API
- LangChain · LangFlow
- Retrieval-Augmented Generation
- LLMs · AI chatbots
- LLM product management
Integrations & platform
- Epic · SMART on FHIR
- Authentication — SSO / MFA
- REST APIs · HL7 FHIR
- Third-party integrations
- SQL · Python (basic)
Product craft
- Strategy · discovery · roadmapping
- Clinical workflow design
- PRDs · specs · success metrics
- Agile · Scrum
- Analytics — GA · Matomo · Mixpanel
Education
MBA, Hospital & Health Management — IIHMR University
Education
B.Sc. (Hons.) Botany — University of Delhi
Certified
PSPO I · SAFe POPM 6.0 · AI Product Management