Page 41 - FCW, November/December 2019

To help maximize the return on federal grant investments, GrantSolutions — a partnership managed by the Department of Health and Human Services — developed Recipient Insight, a cloud-native, scalable intelligent decision engine for risk analysis. It automates the aggregation of publicly available data from a variety of sources, then presents information and analytics in a comprehensive dashboard profile. It pulls data from several sources, including SAM. gov, the IRS, HHS’ Payment Management System and GrantSolutions’ own grant administration and performance data.
Recipient Insight allows users to review key data on over 750,000 potential recipient organizations, using machine learning- powered algorithms that predict outcomes and flag potential performance issues. It has reduced the time to verify key information on more than 50,000 annual grant actions by over 60% and saved more than $5 million annually while freeing staff to focus on higher-value tasks.
Its impact extends beyond the grants management community. In June, Recipient Insight became available to all federal holders of personal identity verification cards, and acquisition leaders are exploring its potential to help evaluate federal contractors.
High Performance Computing Architecture for Cyber Situational Awareness
High Performance Computing Modernization Program, Department of Defense
Since fiscal 2017, the Defense Department’s High Performance Computing Modernization Program has been exploring the applicability and utility of DOD’s HPC resources in the cyberspace domain.
The program provides such resources via the HPC Architecture for Cyber Situational Awareness (HACSAW), which uses cyber data from across DOD’s information networks but especially the Defense
Research and Engineering Network. “HACSAW is designed to ultimately
leverage HPC resources to significantly reduce the time to respond to changes in
the cyber environment from months and days down to minutes,” said Leslie Leonard, a computer scientist at the U.S. Army’s Engineer Research and Development Center. HACSAW’s success could catapult DOD into next-generation cybersecurity capabilities, she added.
The program uses proven open-source technologies — including Apache Kafka, Apache Spark, Elasticsearch, Kibana and Jupyter Notebook — and has focused from the beginning on a purpose-built architecture and processing pipeline to reduce barriers to using real-world data in developing advanced analytics.
The effort kicked into another gear in fiscal 2019, when HACSAW’s high-value dataset began powering the Defense Advanced Research Projects Agency’s Cyber-Hunting at Scale Program. DOD’s Joint Artificial Intelligence Center is also innovating around HACSAW, particularly with the national mission initiative on cyber sensemaking.
Leonard credited Krisa Rowland, the modernization program’s associate director for security, for much of the recent progress. Rowland helped secure some $25 million
in funding for the program, but more importantly, “she’s a connector,” Leonard said. “She knows how to bring people together in and out of DOD.”
Indiana Department of Correction Occupational Optimization Model
Indiana Department of Correction
To ensure that the vocational training that the Indiana Department of Correction had long been providing to inmates maximizes their job potential after release, a team developed the Occupational Optimization Model. It uses statistical, data mining and reporting software to help match soon-to-be- released offenders with job opportunities.
Using data from the Department of Workforce Development — along with education, geographic and corrections information from multiple state agencies
— the model provides a dashboard view of job opportunities filtered by educational requirements, location and pay to show where an offender might find the best employment and wages. Data is presented in several visual formats, such as histograms, pie charts and heat maps. As a user re-sorts the data, the visualizations shift in real time.
The system weights the vocations based on where offenders will live on release, their level of education, the job’s accessibility via public transportation and the potential to earn a living wage. The result is twofold: The model gives corrections officials a high-level look at the vocational training available in the state, where gaps exist and how to address them, and then enables staff to find the most advantageous vocational program for individual offenders while they’re incarcerated. Additionally, parole officers can use the model to determine the best options for released offenders.
“We have always provided vocational training, but...we want to make sure that they have exactly the skills,” said Sarah Schelle, the department’s executive director of legislation and data science. “In the past, we’ve worked on soft skills or not necessarily vocational framing. With the increase in that area, we wanted to make sure we were providing what was actually needed.”
Still in the testing phase, the model will be ready for full use in the next six months at the state’s 18 facilities and 10 parole districts, Schelle said.
November/December 2019
FCW.COM 39