Solace is an AI-powered patient intake and triage application for emergency departments. The application's user-facing components, the patient app and clinician terminal, are hosted on Vercel. These are static, fast-loading shells that cache at the edge, optimizing performance even on weak network connections.The request path initiates with a patient scanning a QR code or a clinician using their terminal, both interacting with the Vercel static sites. These sites then communicate with the AWS backend through CloudFront and WAFv2 for security. An HTTP API Gateway directs requests to a Lambda function.This Lambda function runs a FastAPI service packaged as an ARM64 container image using Python. Trained machine learning artifacts are baked into the container image to avoid slow cold starts. Amazon DynamoDB serves as the primary data store, with over 30 tables encrypted by a customer-managed KMS key.Transient data in DynamoDB utilizes TTL, and the live patient queue is managed via a global secondary index. Media files are stored securely in S3, with all access restricted and delivered via presigned URLs. AWS Bedrock, Transcribe, and Polly handle all AI interactions, ensuring patient data remains within the AWS Business Associate Agreement boundary.Security is paramount, with WAFv2 on CloudFront implementing various security rules and AWS Shield Standard providing protection. Clinicians authenticate via SMART-on-FHIR or a PIN, and IAM follows least-privilege principles with MFA. The overall architecture prioritizes leveraging managed AWS services to enable a small team to deliver a HIPAA-grade, scalable solution. This design choice avoids compromising between development speed and robust functionality.