Amazon Job Advertisement Event System
Serverless microservices platform handling job advertisement lifecycle events at Amazon scale, replacing a brittle monolith that couldn't sustain peak hiring traffic.
Context
Amazon's job advertisement system handled the full lifecycle of job postings across the company's internal hiring pipeline: creation, distribution, updates, and expiration. The existing monolithic application had served the business for years, but scaling it for peak hiring periods had become increasingly difficult. Any change to one part of the system risked breaking unrelated features, and the tightly coupled architecture made it impossible to scale individual components independently.
I joined as a Full Stack Developer through Insight Global and was assigned to lead a 5-developer team tasked with rebuilding the system on a modern serverless architecture.
What I Built
Microservices decomposition: Broke the monolith into independent services, each owning a bounded context of the job ad lifecycle: creation, validation, distribution, status updates, and expiration. Services communicate asynchronously via SQS and SNS, removing the tight runtime coupling that had made the monolith fragile.
Serverless infrastructure on AWS CDK: Provisioned the entire infrastructure using AWS CDK, replacing manual CloudFormation templates with type-safe infrastructure code. All Lambda functions, API Gateway routes, DynamoDB tables, SQS queues, and SNS topics were defined and deployed from a single CDK application, giving the team repeatable, version-controlled infrastructure.
CI/CD pipelines: Implemented fully automated build and deployment pipelines using AWS CodeBuild and CodePipeline. Every pull request triggered an isolated build and test run; merges to main deployed automatically to staging with a manual gate before production.
JWT-based API security: Implemented JWT-based security filters across all backend services, enforcing authentication and authorization on every API call. Replaced the monolith's session-based auth with stateless token validation suitable for a distributed microservices environment.
Secure web architecture: Configured domain registration, Route 53 hosted zones, CloudFront distributions, and TLS certificates for all public-facing endpoints.
Outcome
The rebuilt system handled Amazon's job advertisement workload on a fully serverless foundation. The event-driven architecture allowed individual services to scale independently during peak periods without impacting the rest of the pipeline. CI/CD automation reduced deployment friction and gave the team confidence to ship more frequently.
What I Learned
Leading a 5-developer team on a greenfield rewrite taught me where team autonomy matters most. Keeping service boundaries clear from the start, and owning the data model design before writing code, prevented the coordination overhead that kills microservices migrations. The CDK investment paid off immediately: infrastructure drift stopped being a source of production incidents.
Key Outcome
Transformed a monolithic job ad system into a scalable serverless microservices architecture serving Amazon's hiring pipeline.