AWS Unused Resource Detector

💥 How I stopped burning money on forgotten AWS resources (and built a Lambda to do it for me)

🔍 The Realization

You know that feeling when you check the AWS bill and mutter, "Wait... why is it $600 this month?"

Welcome to the wonderful world of cloud waste — the graveyard of dev/test instances, unattached EBS volumes, and Elastic IPs floating around like lost luggage.

I originally built this while working at AlphaPoint, where we had multiple AWS accounts and environments spinning up and down for blockchain trading experiments. Cloud waste was real.

At one point, we were paying monthly for resources we didn’t even know existed. There was no alert. No dashboard. Just a bill and some confusion.

😅 My Breaking Point

One Monday, I found an EC2 instance labeled alpha-lambda-test-final3 that had been running untouched for 19 days.
The kicker? It wasn’t even mine.

🛠️ My Fix: A Serverless Snitch 🕵️‍♂️

I built a Lambda-based detector and predictor that:

✅ Scans all AWS regions
✅ Flags unused EC2s, unattached EBS volumes, and idle Elastic IPs
✅ Offers optional cleanup (you can delete from the report itself)
✅ Calculates real-time cost impact — so you instantly know how much you'll save
✅ Emails a report and saves a JSON audit log to S3

Unlike AWS Cost Explorer, which takes time to reflect changes, this script gives you a snapshot of your cloud waste and future savings in real time.

💡 Pro Tip

Elastic IPs are only free if they’re attached.
Otherwise… they’re quietly draining your budget.
Ask me how I know.

📬 GitHub Link (You Know You Want It)

This is a cleaned-up, open-source version you can fork, clone, or flex at your next standup:
👉 github.com/chinmaya-chhatre/aws-unused-resource-detector

📈 What Changed

💰 Cut our AWS bill by ~15% in the first week
👨‍💼 Became the unofficial cloud janitor
📊 Finance started sending me thank-you memes. Not even kidding.

⚖️ Tradeoffs I Made

Audit vs. Action: I gave users the option to delete resources but made it manual to avoid unintended deletions. Safety over speed.
Start Small, Scale Fast: The first version focused on EC2, EBS, and Elastic IPs. Once it proved useful, I expanded it to support all major services across all regions.
Cost Approximation: Used general pricing estimates instead of real-time billing APIs — good enough for predictions without adding unnecessary API complexity or latency.

🧠 What I'd Add Next

🔍 Add idle Lambda and RDS detection
💬 Slack alerts for "Why is this EC2 still running?"
🧼 Auto-clean toggle (with approval flow, obviously)

🧵 Why I’m Sharing This

Because cost control shouldn’t feel like wizardry.
Because not every solution needs 6 dashboards and a 12-week roadmap.
Because sometimes, a 100-line Lambda script saves you thousands.
And because no one wants to explain a surprise AWS bill to leadership.

🧠 AWS Unused Resource Detector