Project Overview

Fortuna Helix - Genome Analysis & Prediction Platform

Event-driven serverless system for large-scale genomic data processing

A serverless platform that efficiently processes 100–200GB of genomic sequencing data. AWS Batch and EFS run up to 1,000 samples in parallel, while EFS I/O throttling and Lambda timeout issues were addressed to ensure stable operation.

💡 Key Outcomes

• Faster analysis: AWS Batch + EFS parallel processing cut analysis time from 30 min → 10 min (66% reduction)
• Cost savings: Serverless migration reduced monthly operating cost from ₩6M → ₩500K (92% reduction)
• EFS I/O throttling: Root-cause analysis and a multi-EFS distributed setup to improve throughput
• Lambda stability: Timeout tuning and retry policy design for reliable execution
• PDF reports: SQS FIFO + Lambda generates 1,000 PDFs in about 30 seconds
• Event-driven architecture: EventBridge + SNS/SQS–based pipeline design

🏗 Technical Architecture

1. Compute & Storage

• AWS Batch + EFS: Parallel processing of large genomic sequencing data (100–200GB), 1,000 samples per run
• Lambda: Report generation and orchestration; timeout and retry policies for stability
• S3 · EFS: Raw data and intermediate results; shared storage across Batch jobs

2. Event-Driven Pipeline

• EventBridge + SNS/SQS: Event-based triggers; FIFO queues for ordering and bulk PDF generation
• AppSync · Cognito: API and authentication
• CloudWatch · ECR: Monitoring and container image management

🛠 Key Technologies

NextJS, NestJS, Fast API, TypeScript, Python, AWS Batch, Lambda, Step Function, AppSync, EventBridge, SQS, SNS, EFS, S3, Cognito, CloudWatch, ECR, CloudFormation

Developed with NestJS, AWS Batch, and event-driven serverless architecture.