I help hospitals, health-tech teams, and AI teams turn fragmented clinical workflows, scattered operational data, and frontline documentation into structured systems clinicians can use during real shifts.
Biomedical engineering background with hands-on hospital experience across ICU, laboratory, theatre, diagnostics, and biomedical equipment support.
I work at the intersection of clinical workflows, healthcare operations, biomedical engineering, and AI-assisted systems.
I focus on workflow-first systems that help teams clean healthcare data, map clinical breakdowns, improve escalation logic, and design AI-ready workflows for real clinical environments.
ICU, lab, theatre, and wards each run their own playbook. Handoffs lose context and patients fall through the gaps.
Exports are inconsistent, fields are free-text, and weekly reports get rebuilt by hand instead of generated.
Signals live in separate systems. No one sees the full picture of a patient or a shift in one place.
Critical changes are noticed late. Escalation paths are unclear and SLA timing is not measured.
Key findings sit inside long narrative notes. Nothing pulls them out into a structured, reviewable form.
Models are built without shift realities, escalation rules, or clinician cognitive load in mind.
Charts everywhere, decisions nowhere. Teams still need to translate visuals into next steps.
Tasks float without an owner. Nothing tracks who acted on which signal and when.
A synthetic ICU workflow demo that turns fragmented signals from vitals, labs, alarms, notes, and handoffs into a structured audit trail.
A clinical workflow product concept showing how AI can help clinicians summarize patient context, prioritize tasks, and document actions without adding cognitive load.
Extracts chief complaint, diagnosis, and care plan from free-text medical notes into structured JSON with safety-focused review steps.
ECG classification using MIT-BIH data to detect rare cardiac events across heartbeat classes, with class imbalance handling and model evaluation.
A DenseNet-121 chest X-ray classifier with Grad-CAM heatmaps for interpretability and safe communication of model limitations.
A signal intelligence platform that collects, classifies, scores, and monitors fragmented public signals — an architecture that transfers to healthcare operations and workflow monitoring.
Map clinical breakdowns, escalation gaps, and handoff friction across departments.
Turn messy EMR exports and spreadsheets into structured, reviewable datasets.
Structure, label, and validate clinical data for safe and useful AI workflows.
Specs, workflow diagrams, and review-ready documentation for clinical teams.
Quantifying latency between clinical signals and clinician action across ICU shifts.
Free-text clinical notes to structured JSON with safety-focused review steps.
AI-assisted shift workflow concept that reduces cognitive load without replacing clinicians.
Evaluating ECG and imaging models with class imbalance handling and interpretability.
I am open to healthcare AI roles, clinical workflow projects, AI evaluation work, health-tech product support, and healthcare data systems consulting.