azamalijafri/fintech-data-platform ? reverse-engineered prompt

Build me a local fintech data platform that takes synthetic transaction data, streams it through Kafka, lands it in S3 as raw JSONL files, then lets Snowflake auto load it with Snowpipe and transform it with dbt into clean analytics tables.

I want the whole thing to run with Docker Compose, with a simple setup using a .env file like the example, and a Makefile for common commands. Include a small shared Python package for config and event formatting, a Kafka producer that reads the PaySim CSV and publishes transactions, and a consumer that batches messages and writes them to S3. Then add an Airflow workflow that just runs dbt on a schedule and shows success or failure. Also set up the dbt project with staging, intermediate, and mart models, plus a fraud transactions fact table and basic tests. If anything needs current docs or best practices, look them up online first.