SynfraCore
Synfracore
Start Learning
Navigation

Academies

Platform

RoadmapsLabsCertificationsInterviewPYQsAI AssistantCareer
Start Learning Free🗺️ Learning Roadmaps

Cost OptimizationProjects

Portfolio-ready projects to demonstrate your skills

✍️
Written by senior engineers. Reviewed for technical accuracy.· Updated 2025 · SynfraCore Cost Optimization Team
Expert Content

Cost Optimization — Portfolio Projects

Build these to prove you can actually find and eliminate waste, not just recite savings-plan percentages. Each project maps to a real FinOps practitioner task.

Project 1: Automated Waste-Detection and Cleanup Pipeline

Level: Beginner | Time: 2 days

Build a scheduled Lambda function that scans an AWS account for the classic waste categories — unattached EBS volumes, unassociated Elastic IPs, snapshots older than 90 days, idle load balancers — and reports them (don't auto-delete on a first pass; that's how you accidentally delete something someone needed).

Steps

1.Write a Lambda function using describe-volumes/describe-addresses/describe-snapshots to find each waste category
2.Tag each finding with an estimated monthly cost (EBS $/GB-month, EIP $/hr when unattached)
3.Publish a summary to SNS/Slack on a daily EventBridge schedule
4.Add a --dry-run vs --delete mode, defaulting to dry-run — document why auto-deletion without a grace period is risky (a "temporary" test resource someone's about to need again)
5.Track findings week over week to show whether the waste rate is actually improving

Skills Demonstrated

AWS SDK/CLI resource auditing
Cost estimation from raw resource metadata
Safe-by-default automation design (dry-run before destroy)

GitHub Repo Name

aws-waste-detection-pipeline

---

Project 2: Right-Sizing Recommendation Digest

Level: Intermediate | Time: 3 days

Build a weekly digest that pulls AWS Compute Optimizer's findings, cross-references them against actual CloudWatch utilization, and produces a ranked, dollar-estimated list of right-sizing opportunities — the kind of report a FinOps practitioner actually hands to engineering teams.

Steps

1.Pull findings via aws compute-optimizer get-ec2-instance-recommendations
2.For each OVER_PROVISIONED finding, pull the instance's actual CPU/memory utilization over the trailing 2 weeks from CloudWatch, so the recommendation has real evidence attached, not just Compute Optimizer's verdict alone
3.Estimate the monthly savings for each recommended instance-type change
4.Rank by dollar impact, not alphabetically — the top of the report should be the instance costing the most for what it's actually using
5.Publish per-team (using cost-allocation tags) so each team sees only their own instances, not the whole account's

Skills Demonstrated

Compute Optimizer + CloudWatch integration
Turning raw recommendations into an actionable, evidence-backed report
Per-team cost attribution via tags

GitHub Repo Name

rightsizing-recommendation-digest

---

Project 3: Multi-Account Cost Allocation and Budget-Alert Platform

Level: Advanced | Time: 5 days

Design a cost-visibility platform across a multi-account AWS Organization: enforced tagging, per-team showback dashboards, and proactive budget alerts before a team blows through their monthly allocation — not after the bill arrives.

Steps

1.Define a required-tags policy (Team/Project/Environment/Owner) and enforce it with an SCP or Config Rule that flags non-compliant resources
2.Pull consolidated billing data via Cost Explorer's API, grouped by tag, across all member accounts
3.Build a showback dashboard (even a simple QuickSight or Grafana view) so each team can see their own spend trend without needing account-admin access
4.Set AWS Budgets with alert thresholds (50%/80%/100% of monthly allocation) per team, notifying the team's own Slack channel — not just a central FinOps inbox nobody checks
5.Add a monthly "commitment coverage" report: what % of steady-state usage is covered by Reserved Instances/Savings Plans vs. still running On-Demand unnecessarily

Skills Demonstrated

AWS Organizations + Cost Explorer API at multi-account scale
Tag-enforcement guardrails, not just tag-suggestion documentation
Proactive (before-the-bill) vs. reactive (after-the-bill) cost governance

GitHub Repo Name

multi-account-cost-platform

---

Tips for Great Projects

Make it real. Run these against your own free-tier account first — a waste-detection tool that's never found a single unattached volume in a real account is a tool you haven't actually tested.

Show the dollar figures. "Found 3 unattached EBS volumes" is weaker than "Found 3 unattached EBS volumes costing an estimated $14/month — here's the math." FinOps is fundamentally about making cost visible in concrete terms, and your portfolio project should demonstrate that same instinct.

Explain a real tradeoff. Why dry-run by default? Why per-team dashboards instead of one account-wide view? Being able to explain the reasoning, not just the implementation, is what a FinOps interview actually probes for.

Portfolio Checklist

[ ] All 3 projects on GitHub with clear READMEs showing real (or realistic sandbox) findings, not mocked data
[ ] At least one project outputs an actual dollar-estimated savings figure, with the math shown
[ ] At least one project demonstrates safe-by-default automation (dry-run/confirmation before anything destructive)
[ ] Each project's README explains why the specific cost-governance approach was chosen, not just what it does
Share:
Join our Community
Daily tips, job alerts, interview help — join engineers learning together
Up Next
💬
Cost OptimizationInterview Q&A
Most asked interview questions with detailed answers
Also Worth Exploring
← Back to all Cost Optimization modules
LabsInterview Q&A