Over the previous few weeks, Iāve been constructing one thing Iāve at all times wished to ā a correct fraud detection dashboard that runs on actual information, predicts fraud, and appears like a SaaS product. Itās not a tutorial mission. Itās a full product.
And now itās dwell:
š https://creditcardfraud.info
On this publish, Iāll stroll you thru the stack I used, how I dealt with huge information, the issues I confronted, and a few takeaways. In the event youāre somebody studying ML or backend/frontend dev, that is for you.
The concept was easy: add a CSV of bank card transactions, run predictions utilizing an ML mannequin, and show the outcomes on a clear dashboard.
However when you scale this as much as 1 million+ information and purpose for real-world utilizationā¦ the simplicity goes away in a short time.
