A generalizable application framework for segmentation, regression, and classification using PyTorch

Spatial Multiomics Profiler for Spatial Characterization of Tissue Microenvironment (https://smprofiler.io)

Workflow-centred open-source fully automated lung volumetry in chest CT.

Voice-activated AI medical records assistant using Agno multi-agent framework. Natural language EMR interaction with speech recognition, FHIR-compatible patient chart management & clinical workflow orchestration via FastAPI.

Bridging the gap between theory and bedside practice through open-access, physiologically-weighted, evidence based, decision support tools for anesthesia and intensive care.

Open-source web app that turns patient pre-visit intake into a structured, doctor-friendly summary/PDF (local-first; no diagnosis/treatment/triage).

🩺 Enhance small medical object segmentation with CaraNet, a cutting-edge context axial reverse attention network designed for precise and efficient analysis.

Primary care pre-visit intake summarization MVP that turns structured patient-reported data into clinician-friendly notes with strict safety boundaries.

Nirnay is an AI-assisted clinical workflow application built with Streamlit. It combines patient intake, image support, and a safe medical chat assistant to help clinicians and care teams organize patient data, review clinical images, and explore medical guidance in a polished dashboard.

A thought experiment for healthcare AI: a session-bound clinical reasoning system that can read charts but is architecturally incapable of writing, storing, or remembering medical records.

Administrative decision-support system for prior authorization readiness. Deterministic, rules-first evaluation of documentation completeness with refusal semantics, full auditability, and write-only AI letter drafting. Built for safety, transparency, and regulated healthcare workflows.