Cloud Computing Lets IARPA Carry on During COVID-19

The cloud computing infrastructure at the Intelligence Advanced Research Projects Activity allowed the organization to pivot to a new teleworking norm during the pandemic that’s not much different than the old norm. The organization has conducted business as usual, hiring program managers, adding office directors, creating and killing programs, and continuing to meet the intelligence community’s technology needs.

Catherine Marsh, director of the Intelligence Advanced Research Projects Activity, known as IARPA, was told on March 12 to “lean forward,” and she did, allowing almost the entire staff to telecommute beginning the next day. Even contractors work from home legally, securely and effectively.

“One of the major accomplishments at IARPA has really been our seamless transition to a remote working environment in the cloud,” Marsh says. “That’s because, unlike most of the intelligence community, most of our work is not classified, and we were really prepared ahead of time with an approved, secure, cloud-based virtual environment, and we were able to pivot quickly at having everybody work safely in isolation away from the office.”

She also credits IARPA’s Research and Technology Protection Program, which she describes as a “best practice” that “gives you the clear lanes in the road where you go from unclassified to classified. “It’s a very robust process, and we share it widely, and that allows us to do most of our work in the unclassified world. That is the first stop after you have an approved new [program] start pitch at IARPA.” Marsh reports that her predecessor, Peter Highnam, uses the same program at IARPA’s cousin organization, the Defense Advanced Research Projects Agency (DARPA), which he led as acting director until Victoria Coleman was appointed to be the agency's director.

Fortunately, she says, COVID-19’s impacts on IARPA itself have been minimal. “The disruptions we’ve really encountered relate to challenges our partners are having, doing things safely and securely and things that have been shut down.”

She specifies that some test and evaluation efforts have been delayed. For example, the Little Horned Owl unmanned aerial vehicle (UAV) program was supposed to undergo flight testing in New Mexico in early April, but the flight range had to close. IARPA officials were hoping to continue testing later in summer.

The program is developing quiet, miniature unmanned aircraft technologies to further enable critical intelligence and military missions. Additionally, it will develop acoustic “footprint” prediction tools that are expected to allow operational UAV users to know where they can safely and quietly operate. The program is a continuation of the IARPA Great Horned Owl program, which also researched quiet unmanned aircraft technologies.

“If we are successful with the program results, that’s going to be a game changer for the intelligence community. That program is one of our more classified programs, so I can’t go into much more than that, but these tests are critical,” Marsh asserts. “One of the things you find with unmanned aerial vehicles is that you can hear them before you see them, so that’s an alert, and you don’t want an alert. I’m very excited about it.”

Marsh says her top priority is “continuing to focus on our core value proposition, investing in the high-risk, high-payoff research and development to deliver overwhelming intelligence advantage to the nation.” But to do that, she is making some major changes within the organization, such as bringing back the positions of office directors, which IARPA eliminated several years ago.

The office directors will actively recruit the next generation of IARPA program managers, act as first-line supervisors and help develop and guide research programs. They also will serve as a bridge between program managers and the intelligence agencies and help ensure technologies transition to the intelligence agencies.

It is important to fill those office director positions because IARPA also has been hiring a handful of new program managers. “We look for experts to challenge the status quo and to help shape and structure new programs with metrics and milestones and rigorous testing and evaluation to demonstrate that the proposed solutions will meet intelligence community needs,” she says.

Marsh also is trying to lower the classification level required for program managers to make the hiring process easier. “That will open the aperture even wider and should, hopefully, shorten the timeline for hiring,” she says.

Other changes include eliminating two programs, the Hybrid Forecasting Competition (HFC) and the Homomorphic Encryption Computing Techniques with Overhead Reduction (HECTOR) programs.

The HFC program sought to develop and test hybrid geopolitical forecasting systems by integrating human and machine forecasting components to create maximally accurate, flexible and scalable forecasting capabilities. “With HFC, we were not meeting the metrics, and we had to say these approaches are not going to yield results, and we closed down the program,” Marsh explains. “We don’t stay wedded to ideas here. We kill things that aren’t working, that are not meeting metrics. They weren’t meeting their metrics, and that’s what we have to do. When you take a high-risk, high-payoff approach, that’s okay. It means that not everything you do is going to pan out.”

The HECTOR program is technically dead, but the research will live on in a classified environment. The program’s goal is to enable the development of a broad spectrum of secure distributed applications that use advanced cryptographic techniques. The HECTOR program seeks to develop a comprehensive set of design tools, programming languages and verification tools, according to IARPA’s website.

“With HECTOR, we’re going to pivot into a new direction that has classified aspects to it. For it to transition, it had to have a classified aspect that was missing from the program as it existed. It’s not that the performers were not doing well. They were doing excellent. It’s just the program has to move in a different direction,” Marsh reveals.

Marsh indicates that under normal circumstances, taking over leadership of an organization right before a worldwide pandemic hit would have been a serious challenge, but she had two advantages: she was already the deputy director under Highnam before he went to DARPA, and she had a few months in the position before being forced to create a new normal for the agency. “If I had just come on and they didn’t even know me, I think it would have been impossible, to be honest,” she says.

She stresses the need to “communicate, communicate, communicate” with her staff. “I’m a really big believer that email is not the only way to communicate. In fact, you often have to pick up the phone. When things are going awry, I make sure we do whatever minor course correction needs to happen.”

And there have been a few course corrections, she offers, even if minor. “I can’t say that it’s perfect. There have been some hiccups. I would be remiss if I said everything was perfect.”

One of the ways she communicates is by providing weekly highlights that she also sends to the Office of the Director of National Intelligence. Recent highlights include two programs pivoting to tackle the COVID-19 challenge. The Molecular Analyzer for Efficient Gas-phase Low-power INterrogation (MAEGLIN—pronounced Magellan) program has been developing sensors to detect harmful gaseous chemicals in the air.

The goal is to detect weapons of mass destruction or chemical indicators of illicit activity, such as narcotics production. But now, the program is investigating how well its newly developed micro-gas chromatograph might work as a breath sensor to detect signs of Acute Respiratory Distress Syndrome, a life-threatening condition associated with COVID-19. “It seems that for standoff detection, there’s a unique signature associated with COVID-19, and you can track the progress of the disease, for example, whether it’s waxing or waning, if you will,” Marsh says.

Also, the Functional Genomic and Computational Assessments of Threats (Fun GCAT) program is showing some promise related to the fight against COVID-19. To better address biosecurity concerns, the program intends to develop next-generation computational and bioinformatics tools to improve DNA sequence screening and to augment biodefense capabilities through the characterization of threats.

Those tools are demonstrating high predictive accuracy and a 200 times improvement in computational efficiency over state of the art, Marsh reports. “The Fun GCAT uncovers a variety of different cellular and molecular roles of DNA sequencing and uses that for experimental pipelines. One of those pipelines is testing virus genes for the ability to disrupt the immune system hard-wired within each of our cells. The functions that are coming out of that have been discovered to provide key insights to the threat of the novel COVID-19 virus,” she adds.

Whether during the old, pre-COVID normal, or the new, Marsh credits IARPA’s high-risk, high-payoff philosophy for the organization’s success. “It’s that value proposition that we offer for the community, thinking that we can take that high-risk approach and we can fail, and that’s okay. That’s a different approach to things than others might be willing to take on, but that’s how you make a difference.”