top of page
  • Medium
  • LinkedIn
  • RStudio
  • githubgrey
  • kaggle
  • X
  • watt
  • Instagram

Custom Fraud Prevention Solutions

Fraud Prevention Solutions

Utilizing your data we create a 360 view of the company and alert you to threats with a severity and likelihood score

Internal Fraud Solutions

Fraud hurts.  It particularly hurts when the sources are internal. There are many reasons internal fraud occurs, from dishonesty ,to culture, to misapplication of policy or procedure.  Fraud costs.  With our solutions we get to the root cause quickly.

Replicable & Deployable

We have several engagement models to choose from - from data diagnosis and flash modeling, to deployment of the full fraud prevention solution -  Let's talk!

Use Cases

Behavioral Data Science & Fraud Prevention

Internal Theft of HSA's

A health care insurance company was about to face an audit.  Most of their data was in disparate systems, from spreadsheets, to SQL databases, to Access.  Taking call-center records, and applying specially built text analytics libraries, as well as utilizing segmentation and anomaly detection, accounts were located where internal theft of elderly person's accounts were occuring.  A model was deployed to prevent future occurrences.

Social Media Signals Alert Stock Take Over Attempt

A social media reputation analysis model for a cosmetics company in several languages, alerted the company to a stock take over attempt, which would have had serious consequences.

Redefining Fraud to Prevent Future Cases

A financial institution had been defining fraud as "zero pay" - or default.  Meeting with stakeholders from several departments, a new definition of a "Fraud," target was reached.Taking cases from investigators, querying the SQL database, and applying text analytics to call center records started the sampling. Two models emerged, one identifying a pattern of identity theft, which could be resolved quickly, resulting in a fast win.

Forensic Investigation of Accreditation Code

An urban Midwestern University was preparing for their accreditation.  The two programmers who built the code for the accreditation scoring obtained access to staff and faculty salaries.  When their demands for raises weren't met, before quitting, they destroyed the accreditation code. Going in as a consultant, and accessing a little known IBM "back-door" file most of the code was retrieved as well as evidence obtained that the destruction was intentional.

Financial Fraud

Leading a team of enterprise architects and programmers, overseas, the problem to be solved was a specialized form of currency which was being used fraudulently.  With SME's from law enforcement, web-scraping from social media, and other sites, a machine learning model was built that used language as a marker where persons knew to turn off geolocation and encode their transactions in slang

Bank Vault Networking and Integration Issues

A small branch was demonstrating anomalous transactions.  On-site examination showed that regulatory practices were not being followed correctly.  Additionally the cash machines were not properly networked and the software reading transactions was not configured correctly giving erroneous readings and room for dishonest behaviors.

"I have known Dawn since 2013 when we worked together on the Watson Social Media Analytics team. Dawn's background in psychology and statistics gives her unique insight into building linguistic assets. She led an IBM technical team of enterprise intelligence experts that built a well crafted situational awareness solution that improved law enforcement outcomes and actually saved lives. I wholeheartedly recommend her deep subject matter expertise and computational skills as a data scientist"

Brian K, Ex-Googler &

Cloud Solution Architecht

Some of Our Clients

szz.jpeg
ey.png
truven.jpeg
wtwsmall.png
ufcsmall.png
download.png
ATB.png
disneyla.png
uofc.png
interac.png
usaa.png
VA.png
peds.jpeg
ibo.jpeg
jandj.png
coca.png
ATT.png
epic.png
waynestate.png
uhg.png
bottom of page