Real-Time Shop Data Streaming

01 Context

A pipeline simulating shop sales in real time: tickets are generated continuously, flow through Kafka, get stored in PostgreSQL and are exposed through APIs. The whole platform starts with a single docker-compose command.

02 Problem

Real-time sales data is only valuable if the backbone never drops events, the services are reproducible to run, and the team can see stock levels, per-store sales and low-stock situations as they happen.

03 What I built

1. 01

   Kafka backbone - the ingestion server receives tickets and publishes them to Kafka.

2. 02

   Storage - PostgreSQL for tickets, stock and alerts.

3. 03

   Orchestration - docker-compose for every service (Zookeeper, Kafka, PostgreSQL, apps, monitoring).

4. 04

   Observability - Prometheus and Grafana.

5. 05

   Admin API - exposes tickets, sales per store, current stock and low-stock alerts (<10% threshold).

04 In action

Screenshots from the project repository.

05 Results

• sales pipeline on a Kafka backboneReal-time
• stock threshold triggering alerts<10%
• docker-compose up to run the platform1 command

06 My part

07 What I learned

• Kafka decouples producers from consumers: ingestion keeps flowing even when storage or alerting services lag behind.
• docker-compose turns a multi-service platform into something anyone can start with a single command.
• Instrumenting a real-time system from day one (Prometheus, Grafana) is the only way to see what it is actually doing.
• Clear boundaries between platform and application services make a two-person streaming project manageable.