← Back to Case Studies
TD Bank Group

A fraud dashboard for one card business

Oct 4, 2026Capability study
Data & AnalyticsFintech

Services Provided

Data & AnalyticsAI & Machine Learning

Tech Stack

PythonSupabaseNext.jsTypeScriptAzure

Overview

Airodevs built a fraud-monitoring dashboard for one card-issuing business line of a large Canadian bank, giving fraud analysts a single live view of flagged transactions instead of the daily batch export they had worked from before. The project ran under the bank's own data-governance and OSFI-aligned risk review process, with all data held in Canada.

Client

The client is one of Canada's largest banks, with a card-issuing business line processing millions of transactions a day. Its fraud team monitored flagged transactions through a daily batch report, built and reviewed under the bank's model-risk governance ahead of any change going live.

Challenge

Fraud analysts worked from a report generated once overnight, so a fraud pattern starting mid-morning was not visible until the next day's file. Analysts also had to cross-reference three separate systems, card type, merchant category and customer history, to judge whether a flagged transaction needed a call to the cardholder.

Solution

We built a dashboard that pulled flagged-transaction data on a near-live feed instead of the overnight batch, merged card type, merchant category and customer history into one record per flag, and ranked flags by a risk score the fraud team's own model already produced, so the dashboard added visibility without introducing a new scoring model into the bank's risk review.

Areas of Impact

The fraud team's intraday monitoring, which moved from a next-day batch report to a near-live queue; analyst case review, now built around one merged record per flagged transaction instead of three separate system lookups; and the time to first contact with an affected cardholder on confirmed fraud cases.

Results

83%
faster time to flag review

review start time down from 4.2 hours -> 43 minutes

29%
more confirmed fraud caught same day

same-day catches up from 61% -> 79% of confirmed cases

3 systems
merged into one flag record

lookups per case down from 3 -> 1

More Case Studies

Project inquiry

Tell us what needs to work better.

Share the workflow, product problem, or delivery pressure you want to improve. We'll respond with a clear next step.

What happens next

  1. Airodevs reviews the request.
  2. A short discovery call confirms scope and constraints.
  3. A concrete next step or proposal follows.
Office

Toronto, Ontario
Canada

Send us a message

Fill in the details below and we'll get back to you within 24 hours.