project catalogue
Cloud Architecture2026Completed

SLA Monitor

A serverless AWS platform that monitors endpoints minute by minute, detects incidents, and generates weekly SLA reports.

scroll
Type
Cloud Architecture
Role
Solo Project
Status
Completed
Year
2026
The Backstory
AftertheURLshortenerIsetmygoalonamoreambitiousproject,andaprojectthatwoulddeepenmyunderstandingofCloudArchitectures,andmyengineeringapproachandproblemsolving.
Ibuiltitintheordertheproblemitselfdemanded.FirsttheMonitorLambda:EventBridgewakesiteveryminute,itprobeseachtarget,andlogsstatusandlatencytoDynamoDB.
OnlyoncethathistoryexistedcouldIreasonaboutthenextquestion.Onefailedpingisn'tanincident,threeinarowis,sotheSLAProcessorcamenext,runninghourly,openingincidentsonconsecutivefailuresandclosingthemthemomentatargetrecovers.
TheReportGeneratorcamelast,becauseitneededtheothertwotoalreadybetrustworthy:everyMondayitrollstheweekintouptime,p50/p95/p99latency,errorrate,MTTRandMTBF,checkseachnumberagainstthatproject'sownthresholds,andmailstheverdictoutthroughSES.Five
Lambdas,fivedifferentclocks,eachdoingexactlytheirjob.
TheAPIandauthlayercamelast.Cognitohandlessign-inwithGoogleOAuthwiredinasanIdentityProvider,APIGatewayvalidatestheJWTbeforeanythingreachescompute,andtheread-onlyAPI'sLambdaroleisscopednarrowenoughthatevenafullycompromisedendpointphysicallycannotwritearecord.EveryresourcebehindallofitwentinthroughTerraformmodules.

— walid, 2026

Under the Hood
  1. Five Lambdas, each its job

    Monitor every minute, incident detection every hour, reports every Monday, a read-only API, and a project manager for writes. each function triggered on its own schedule with EventBridge.

  2. Professional Metrics

    Weekly reports compute uptime, p50/p95/p99 latency, error rate, downtime, MTTR and MTBF, evaluate them against per-project thresholds, and assign a severity before landing in user's inbox.

  3. Terraform Provisioning

    Clear Modular IaC with remote state in Terraform Cloud. Lambda deployments are content-hashed and immutable, a code change produces a new artifact, and nothing was made in the console.

  4. Least-privilege access control

    API Gateway validates Cognito JWTs before any request reaches compute. Each Lambda function runs under its own scoped role, no shared roles, no wildcard ARNs. If the read path were ever compromised, not a single record would ever be written.

In the Wild
SLA Monitor sign-in secured by Amazon Cognito

Signing in runs through Cognito, and if you'd rather not make a password you can just use Google.

SLA Monitor dashboard with monitored endpoints

The home view. Every endpoint I'm watching in one place, each with its health, latency, and anything going wrong right now.

SLA Monitor endpoint detail with latency analytics and vitals

Opening a single endpoint gives you the last 24 hours of latency, plus its uptime, when the certificate expires, and the most recent checks.

SLA Monitor reliability metrics and SLA report history

The reliability panel works the numbers out for you: MTTR, MTBF, error budget, the incident log, and every SLA report generated so far.

SLA Monitor generating a fresh on-demand SLA report

Press generate and a new report is ready in a second or two, already scored and graded.

Exported HTML SLA report for a monitored endpoint

The report that actually gets emailed out. A clean page on its own, graded against the targets set for that endpoint.

AWS architecture diagram of the monitoring system

The whole system in one diagram: five Lambdas running on five different schedules, and exactly what each one is allowed to touch.

swipe →

Applied Technologies
See It Yourself

The code is public. If you'd rather see it running, request a demo and it lands straight in my inbox.

end of case study — SLA Monitor

Like what you saw?
Let's build something together.

I'm open to opportunities, collaborations, and conversations

if this project sparked an idea, my inbox is the shortest path to it.

browse the full catalogue