About CrossCures
We are building the patient memory layer that clinicians can trust—every claim cited, every gap flagged.
Our Origin
CrossCures was founded by a team of clinicians, AI researchers, and platform engineers who witnessed firsthand the inefficiencies and risks of fragmented patient data. Clinicians spend hours piecing together patient histories from disparate EHR systems, often missing critical context buried in unstructured notes.
Existing AI summarization tools promised to help, but their black-box outputs created new problems: clinicians couldn't verify claims, liability concerns mounted, and trust eroded. We realized that the missing ingredient wasn't more AI—it was provenance. Every clinical claim needs a source citation. Every gap in the data needs to be flagged, not hidden.
CrossCures was built to solve this problem. We combine foundation models with rigorous citation tracking and uncertainty awareness to deliver patient memory that clinicians can trust. Our partnerships with leading health systems like Mass General Hospital, Mass General Brigham, and Mayo Clinic validate our approach and guide our roadmap.
Our Team
Shaurjya Mandal
Technical Lead
MS CS
Foundation-model healthcare AI specialist. Leads model development and deployment infrastructure for CrossCures.
Focus: Model + Deployment
Kapil Dev Nayar
Clinical Lead
MD, Internal Medicine
Internal Medicine clinician with expertise in clinical AI research. Guides workflow integration and clinical evaluation.
Focus: Workflow + Evaluation
Nitin Pasumarthy
Platform Lead
Staff Applied ML Researcher (LinkedIn)
Staff Applied ML Researcher at LinkedIn. Leads platform architecture for secure memory and audit logging systems.
Focus: Secure Memory + Audit Logging
Akshatha Nayak
ML Lead
PhD Bioinformatics (UT Austin)
PhD in Bioinformatics from UT Austin with focus on AI in Medicine. Leads model training and reasoning evaluation.
Focus: Training + Reasoning Eval
Our Values
Provenance First
Every clinical claim must be traceable to its source. No black-box assertions.
Clinician-Centered
We build for clinicians, not around them. Workflows must integrate, not disrupt.
Honest About Uncertainty
Missing or ambiguous data is flagged explicitly. We don't hide gaps or hallucinate facts.