Not in a demo. Not in a pilot. In the real workflow of a real clinician, seeing a real patient. That question shapes everything we build.
I am a physician and clinical informatician building toward a clearer answer to a hard question. What does it take to make AI work at the bedside, not in a benchmark.
As CEO of Krishmatics, I am building a clinician-designed medical SaaS platform with AI-assisted documentation and prescription tools already live in India, and a US release in progress.
As Managing Partner at Anuvansh, I lead a US and India-based practice that helps healthcare organizations design FHIR-native architectures, integrate genomic and clinical data, and operationalize AI in ways that withstand clinical, regulatory, and ethical scrutiny.
My focus sits at the intersection of three things. Clinical LLMs that are grounded, evaluable, and safe enough to trust at the bedside. The precision-health stack, from variant interpretation to multimodal integration of EHR, lab, imaging, and genomic data. And the boring but decisive plumbing, FHIR, OMOP, SMART on FHIR, CDS Hooks. The rails that determine whether any of this actually reaches a patient.
Most of what we build sits in the gap between clinical AI hype and clinical AI that survives contact with a real patient.
Grounded in verified sources, evaluable with real-world test sets, and safe enough that a physician can rely on them in a consult — not just impressive in a benchmark.
Variant interpretation, pharmacogenomics, and multimodal integration of EHR, lab, imaging, and genomic data. The full picture of a patient, not a single data type.
FHIR, OMOP, SMART on FHIR, CDS Hooks, Epic and Cerner integration. The boring but decisive rails that decide whether any of this actually reaches a patient.
Every product decision goes through clinical review before it becomes code. The default answer to "should we build this?" is no, unless a clinician says yes.
We design for the consult that is already happening, not for the consult we wish was happening. Tools live where the work is, not where the demo is.
DPDP-aligned from day one. Encrypted at rest and in transit, India data residency, consent-based access, no model training on patient data.
Production-grade engineering. We earn trust through software that works in the real clinic, not through claims about what it could one day do.
Clinical depth, informatics leadership, and senior advisory, in the same founding team.
Physician and clinical informatician focused on making AI work at the point of care, not just in a demo. Led informatics and data science within a $950M federal HIV prevention portfolio, and is Managing Partner at Anuvansh, designing FHIR-native architectures and operationalizing clinical AI. Published across JAMA, JAMIA, JMIR, and AIDS.
Owns all drug-interaction logic, safety guardrails, and alert rules across the platform, the clinical depth that every competitor in this space lacks. PhD in Health Sciences with a specialization in Health Leadership from Rutgers University, and an MPH in Epidemiology from Emory University.
A widely respected leader in health technology and data science. IIT Kanpur, with an MS and PhD from UC Berkeley. Former Chief Scientist at the Qatar Computing Research Institute, data mining architect at Amazon, and CTO at Persistent Systems. Over 100 invited talks across more than 35 countries. Advises on platform direction, architecture, and long-term positioning for Ruby Rx and Ruby AI.
We are open to senior advisory, board, and select operating conversations in clinical AI, health data infrastructure, and precision health.