I build AI systems that work in real hospitals — and try to make sure they're safe enough to be there.
About
I'm a data scientist at Dr. Lal PathLabs, where I work on deploying machine learning in diagnostic workflows — the kind of work where getting it wrong has real consequences. I'm also a research assistant at the VEDAs Lab, MNNIT Allahabad, where my focus is computational pathology and medical imaging.
In 2025, I co-founded the Foundation for Ethical AI in Healthcare (FEAIH) — India's first independent body for AI standards and accreditation in clinical settings. The idea came out of a recurring frustration: AI systems performing brilliantly on benchmarks, then failing quietly in production, with no one asking why or who was accountable.
I'm particularly interested in federated learning for healthcare — getting models to learn from distributed hospital data without centralising sensitive records — and in the harder question of what it actually means for a medical AI system to be "safe enough" to use on patients.
My academic background is in data science (MSc, BITS Pilani). My practical education has mostly been in the gap between what the literature promises and what real systems deliver.
Research Focus
My work sits across four connected areas — all of them concerned, in some way, with the question of when it is actually safe to deploy AI in clinical settings.
Applying deep learning to whole-slide histopathology images — building models that assist pathologists with grading, segmentation, and anomaly detection. Interested in the clinical validation gap between benchmark accuracy and real-world diagnostic value.
End-to-end model development for diagnostic imaging — from data pipelines through training to deployment. Focus on robustness across scanner types, patient demographics, and clinical settings that differ from training data.
Training machine learning models across distributed hospital systems without moving sensitive patient data. Interested in the practical barriers — heterogeneous data, institutional incentives, communication overhead — not just the theoretical guarantees.
The regulatory, ethical, and institutional frameworks that determine whether AI systems actually improve patient outcomes at scale. This is why I started FEAIH — the technical work is incomplete without the governance layer.
Experience
Co-founded India's first independent standards and accreditation body for clinical AI. Leading the development of the Indian Healthcare AI Standards (IHAS) framework, engaging with regulators, healthcare institutions, and AI developers to build a credible accreditation pathway.
Building and deploying machine learning systems in production diagnostic workflows. Work spans model development, validation, and the practical challenges of keeping AI systems reliable and fair across diverse patient populations.
Research in computational pathology and medical imaging AI. Contributing to projects on deep learning for histopathology analysis, with a particular focus on model robustness and clinical validation methodology.
Graduate training in statistical learning, machine learning, and applied mathematics. Thesis work focused on applications in healthcare data.
Projects
India's first independent standards and accreditation body for AI in clinical settings. Developing the IHAS framework — a three-tier pathway for evaluating healthcare AI systems on safety, fairness, and accountability.
Visit feaih.org →End-to-end deep learning system for whole-slide image analysis in histopathology — covering ingestion, tile extraction, model inference, and structured reporting. Built for deployment in real diagnostic labs.
Inquire →Research into practical federated learning for multi-site healthcare AI — addressing heterogeneous data distributions, privacy constraints, and the governance questions that make multi-institution collaboration complicated.
Inquire →Publications
Research is ongoing. Publications from my work at VEDAs Lab and FEAIH will be listed here as they become available. If you're interested in collaborating on research, reach out directly.
Writing
I'm working on a series of pieces about clinical AI deployment in India — the regulatory gaps, the real-world failure modes, and what governance infrastructure needs to exist before we can deploy AI at scale in healthcare. Check back, or reach out if you want to talk through any of these ideas before they're written up.
Contact
I'm open to research collaborations, speaking on healthcare AI and governance, consulting on clinical AI deployment, and conversations with anyone thinking seriously about how AI should work in medicine. No cold pitch decks, please — a genuine question works better.
If you're working on something in the space of medical AI, federated learning, or healthcare AI governance and want to think through it together — I'm interested. Write with some context on what you're doing.
Write to Me →