Spark + AI Summit 2019 Taps Kount’s AI Science Team for Best Practices in Machine Learning and AI-Driven Results

Kount Reduced Time for MLflow Model Training by 87 Percent and MLeap
Execution by 65 Percent

BOISE, Idaho–(BUSINESS WIRE)–Kount, a leading fraud prevention solution, today announced that two of
its team members will speak at the Spark + AI Summit, taking place April
23-25, 2019 in downtown San Francisco. The event is the largest data and
machine learning conference in the world and brings together developers,
data engineers, data scientists, and decision-makers to collaborate at
the intersection of data and machine learning. Josh Johnston, Director
of AI Science at Kount, and Noah Pritikin, lead Site Reliability
Engineer for the AI Team, will discuss how Kount uses artificial
intelligence and machine learning to protect the digital innovations of
6,500 customers from digital fraud.

Johnston speaks at 1:40 p.m. on Wed., Apr. 24, and will describe how
Kount cut model training time from two days to two hours, reduced failed
runs, and tracks experiments better with MLflow. Ultimately, moving
Kount to the Spark ecosystem converted an ad hoc and hands-on training
process into a fully repeatable pipeline that meets regulatory and
business goals for traceability and speed.

“Fundamentally, science must be repeatable. The tools and processes we
use for performing machine learning at Kount ensures best practices of
rigorous scientific inquiry,” said Johnston.

Pritikin takes the stage at 3:20 p.m. on Wed., Apr. 24, to discuss
Kount’s real-time architecture for machine learning execution with
MLeap. His talk describes a way to bring a machine learning model that
was trained in Apache Spark for use in production. This architecture
provides reliability, scalability, traceability and model governance
throughout the entire lifecycle while cutting execution time by nearly

“At Kount, we provide our customers with innovative machine learning
solutions that have ‘real-time’ requirements. In order to accomplish
this, we created a reliable, yet scalable, architecture that can deliver
fraud predictions in less than 20ms, 99% of the time. A core component
of our architecture is MLeap,” said Pritikin. “It enables us to deploy
Spark-generated models into production while delivering on our
customers’ latency expectations.”

The Kount team’s participation at the Spark + AI Summit demonstrates the
fraud prevention solutions’ leading and innovative use of AI and machine
learning. Kount employs both supervised and unsupervised machine
learning to provide its global customers with advanced fraud prevention
to protect against payments fraud, account takeover, and new account
creation fraud. This technology combines with a data network with 12+
years of detail, a flexible policy engine, and self-service analytics.

About Kount: Kount’s award-winning digital fraud prevention
solutions are used by 6,500 brands globally, helping them to reach their
digital innovation goals. Kount’s patented technology combines device
fingerprinting, supervised and unsupervised machine learning, a robust
policy and rules engine, self-service analytics, and a web-based
case-management and investigation system. Kount’s solutions stop fraud
and increase revenue for digital businesses, acquiring banks, and
payment service providers.


Jenny Overell

Sky Optics Media drone video