MBMadhav Bhasininblog.madhav.dev·3d ago · 10 min readObservability and Scaling: Monitoring Event Lag and Handling FailureA distributed event-driven pipeline fails differently from a monolith. When a single service goes down, you get an error page. When an event-driven pipeline degrades, it's often invisible at first — m22J
MBMadhav Bhasininblog.madhav.dev·5d ago · 10 min readProduction Infrastructure: Terraform, Autoscaling, and Zero-Downtime DeploysBy Day 5, every layer of PulseCart's event-driven pipeline exists as application code. The FastAPI producer, Cloud Run consumers, Cloud Tasks queues, and Airflow DAGs are all written and tested. What 00
MBMadhav Bhasininblog.madhav.dev·Jul 5 · 8 min readOrchestrating Batch and Scheduled Workflows with AirflowBy Day 4, PulseCart's reactive layer is fully operational. Events flow from the FastAPI producer into Pub/Sub, Cloud Run consumers react in real time, and Cloud Tasks handles delayed work like cart ab00
MBMadhav Bhasininblog.madhav.dev·Jul 1 · 8 min readCloud Run Consumers and Cloud Tasks: Reactive and Delayed WorkflowsIn Day 3 we built the FastAPI producer that publishes events to Pub/Sub. Now we build what consumes them. This is where the event-driven model pays off — multiple independent services reacting to the 00
MBMadhav Bhasininblog.madhav.dev·Jun 28 · 7 min readFastAPI Producer: Publishing Commerce Events to Pub/SubWe have a taxonomy (Day 1) and a Pub/Sub backbone (Day 2). Now we build the service that connects them: the FastAPI ingestion service that receives commerce events, validates them, and publishes them 00
