How I Built an Open-Source AI AWS Cost Optimizer at 16

AWS bills are the silent killer of AI startups. Most founders are burning 30% of their runway on idle compute and orphaned storage without even knowing it.

Seeing how inefficient current FinOps tools are, I spent my weekend building FinOptic — a hyper-minimalist, open-source AI cost optimization engine. I'm 16, and I love building high-performance infrastructure.

🔥 Core Features

The TypeScript backend instantly filters the billing noise and isolates 3 critical drains:

1. Underutilized EC2 Instances: Catches idle compute (CPU < 5%, costing >$200/mo).
2. Orphaned Storage Purge:Locates detached EBS volumes billing hourly in an available state.
3. Wasteful AI Inference Scanner:Flags expensive idle GPU workloads (p3/p4/g5 instances) running 0% active workload.

📊 Performance Benchmark

I stress-tested it using an official production dataset from AWS Labs with 348 active enterprise instances. In less than 5 seconds, FinOptic located hidden GPU infrastructure waste, saving thousands of dollars annualized.

Best part? It is 100% private. All telemetry & security settings are entirely saved in browser localStorage — zero data leaks.

Let's stop overpaying Amazon. Stars and Pull Requests are welcome!

👉 GitHub Repository: https://github.com/soltanovicdana-web