Designed and built a full-stack medical data platform that analyzes large volumes of ECG signals to detect cardiac arrhythmias using machine learning.

The system processes large-scale time-series ECG data streamed directly from medical devices, runs automated arrhythmia classification using ML models, and provides a specialized interface for technicians to review, validate, and retrain predictions.

I was responsible for building the core architecture connecting device data ingestion, ML inference services, and technician workflows, working closely with AI researchers, annotation teams, product managers, and designers.

Built both the frontend and backend systems powering the ECG analysis workflow.

Developed the frontend foundation for rendering high-resolution ECG time-series data, enabling technicians to analyze cardiac events directly in the browser.

Designed backend infrastructure that receives raw ECG streams from medical devices via webhooks, queues incoming data, and processes it through ML inference services.

Implemented Python services responsible for ingesting ECG data and triggering arrhythmia classification models built by the ML team.

Built interfaces allowing technicians to review predictions, annotate events, and feed corrected data back into the ML pipeline to improve model accuracy.

Acted as the integration point between AI researchers, annotation teams, and product stakeholders - defining data structures, communication channels, and system contracts so all components worked seamlessly.

• Architected backend pipeline for streaming ECG signal ingestion • Built Python inference APIs integrating ML arrhythmia models • Designed queue-based processing architecture for device data ingestion • Implemented browser-based ECG waveform visualization for clinical review • Developed technician tools for annotation and model retraining workflows • Defined cross-service data schemas used across AI and application systems

React • Next.js • TypeScript Node.js • NestJS Python • Flask Redis AWS (EC2, S3)