The problem GenomeGuard solves

Today, understanding genetic disease risk is difficult, expensive, and often unsafe from a privacy perspective. People usually discover life-threatening genetic conditions like cancer, Alzheimer’s, or cardiovascular disorders after symptoms appear, when treatment becomes more complex and costly. Traditional genetic analysis requires sending DNA data to external labs, creating data privacy and security risks, long wait times, and high interpretation costs.

At the same time, raw genome files (VCF) contain millions of variants that are impossible to interpret manually without advanced bioinformatics and clinical expertise, making meaningful personal genomics inaccessible to most individuals, hospitals, and researchers.

GenomeGuard solves this by providing fast, automated, private, and AI-powered genomic risk analysis—right on the user's trusted environment (local/cloud), without sending DNA anywhere. It turns complex genetic variant data into clear disease-risk insights, empowering early detection, preventive healthcare, and precision-medicine workflows while ensuring 100% data control and privacy.

Challenges we ran into

During the development and deployment of GenomeGuard, I faced several technical hurdles. The biggest challenge was getting the backend API to properly route through CloudFront. While the API worked when accessed directly via the Application Load Balancer, CloudFront initially returned HTML instead of JSON, causing 502/504 errors. I solved this by debugging path-based routing, updating cache behaviors for /api/*, fixing ECS security group rules, and running CloudFront invalidations until the behavior propagated. Another challenge was ECS tasks repeatedly failing health checks due to an incorrect health-check path and blocked port traffic, which I resolved by adjusting ALB target group settings and updating security group rules. I also had to troubleshoot secure connectivity between ECS and DocumentDB in private subnets, ensure IAM task roles and Secrets Manager access worked correctly, and fix a CI/CD issue where the wrong Docker image was being pushed due to cached build layers. Lastly, I standardized JWT environment variables to ensure consistent authentication across local and cloud environments. These challenges helped me strengthen my understanding of cloud networking, distributed debugging, container orchestration, and secure architecture in real-world deployments.