Our Programs

With its community-wide charter since its inception, IARPA has introduced approximately 90 research programs in diverse areas, including quantum computing, neuroscience, cognitive psychology, sociology, power sources, antennas, as well as chemical and biological sensing. We are at the forefront of leveraging AI and machine learning (ML) to develop remarkable speech and facial recognition capabilities, as well as machine translation and information discovery. For example:

  • Launched in 2011, IARPA’s Babel program developed agile and robust, rapid speech recognition technology that can analyze any human language in order to help analysts effectively and efficiently process massive amounts of real-world recorded speech. Babel focused on underserved languages, such as Pashto, Tamil, Igbo, and others, due to USG partner interest in regional emergent threats. Also, by selecting languages with very grammatical systems, the program was better able to assess performer systems in a wide variety of scenarios. Speech data was made available through the Linguistic Data Consortium (LDC).

    While the technology Babel developed has significantly improved since the program closed in late 2016/early 2017, the Babel team and LDC still receive requests for the data from USG partners. The program’s primary impact is the datasets it created, which are famous in the community, as well as the development of Kaldi, a widely-used, open-source speech recognition toolkit.
  • The Open Source Indicators (OSI) program was introduced in 2011; a team involved with this program was the first to notify U.S. public health officials about the 2014-2016 Ebola outbreak in West Africa.
  • The High Frequency Geolocation (HFGeo) program, which began in 2011, developed a capability that dramatically improved the USG’s ability to geolocate and characterize high-frequency (HF) emitters. Some key accomplishments include: an integrated system that significantly improved HF signal geolocation accuracy; a successful field demonstration; and the transfer of HFGeo-developed technology to government partners. The HFGeo team, which was led by former PM Torreeon Creekmore, was awarded the prestigious DNI Science and Technology Award for their groundbreaking research.
  • Signal Location in Complex Environments (SLICE), HFGeo’s classified sister program, launched in 2011 and focused on enhancing geolocation in complex environments, primarily from long standoff receivers. The challenges addressed include low signal power, emitter motion, multipath propagation, and dense interference environments. The SLICE team received the DNI Team award for its efforts.
  • Launched in 2010, IARPA’s Multi-Qubit Coherent Operations (MQCO) program, which aimed to resolve the technical challenges involved in fabricating and operating multiple qubits in close proximity, was fortunate to have the 2012 Nobel Prize Laureate in Physics, Dr. David Wineland, working as a researcher on the program. The program’s end goal was to execute quantum algorithms using multiple qubits and to evaluate the performance using a metric that can scale to higher qubit numbers.
  • The Great Horned Owl (GHO) program, which launched in 2012, greatly enhanced the Intelligence, Surveillance, and Reconnaissance (ISR) capabilities of unmanned aerial vehicles (UAVs). GHO ended in 2014 after successfully demonstrating a quiet propulsion system for UAVs with more endurance and payload capability. This system quietly generates electrical power from liquid hydrocarbon fuel (specifically gasoline or diesel) and enables purely electrically-driven quiet flight. In fact, an IARPA team, with Air Force Research Laboratory (AFRL) and NASA support, flew the battery GHO UAV (XRQ-72B) on Edwards Air Force Base's dry lake bed in October 2018. Special Operations Command officials were so impressed with how quiet the UAV was that it led to DARPA’s Series Hybrid Electric Powered AircRaft Demonstration (SHEPARD) program.
  • IARPA’s Sirius program, launched in 2012, was the first program to address cognitive bias mitigation training using Virtual Learning Environments (VLEs) that produced validated cognitive bias assessment measures. Sirius has been IARPA’s most transitioned and inquired about program, with over 20 transitions and counting.
  • IARPA’s Janus program dramatically improved facial recognition software performance by increasing identity matching speed and accuracy. Launched in 2014, Janus’ goal was to revolutionize face recognition by fusing information available from multiple views from diverse sensors and visual media to deliver dramatic improvement in speed and accuracy. Janus’ accomplishments include, among others, producing algorithms twice as accurate as the most widely used government-off-the-shelf systems and achieving 85% image verification accuracy at a false match rate of 1 in 100,000.
  • Launched in 2015, the Trojans in Artificial Intelligence (TROJAI) program, aimed to defend AI systems from intentional, malicious attacks, known as Trojans, by conducting research and developing technology to detect these attacks in a completed AI system. Several performer teams who worked on TrojAI also participated in the NeurIPS Trojan Detection Challenge, which invited participants to detect and analyze Trojan attacks on deep neural networks that are designed to be difficult to detect. The Purdue-Rutgers team placed second in the primary rounds for "Target Label Prediction" and "Trigger Synthesis," while the Peraton IUB team placed first in the final round of the competition.
  • Launched in 2016, the Standoff Illuminator for Measuring Absorbance and Reflectance Infrared Light Signatures (SILMARILS) program aimed to develop a portable system for accurate real-time standoff detection and identification of trace chemical residues on surfaces using active infrared spectroscopy at up to a 30 meter range. By the time SILMARILS closed in 2021, the program had achieved a number of impressive results, including: detecting explosives on portable electronics; detecting trace quantities of narcotic simulants through a plastic bag; and detecting target chemicals on a wide range of “wild” substrates with real world clutter, among others.
  • In 2017, IARPA released the research results and forecasting data generated by its Aggregative Contingent Estimation (ACE) program, which, when launched in 2011, initiated a massive competition to identify cutting-edge methods to forecast geopolitical events. This included millions of participant forecasts made over four years of the program’s execution, which led to critical insights into the accuracy of human judgement about geopolitical affairs and aggregated vs. individual forecaster performance. The clear winner from this effort was Team Good Judgment, which went on to build the forecasting business, Good Judgment. In addition, the principal investigator, Philip Tetlock, wrote a popular book, Superforecasting, based on this effort. The IC prediction market preceded ACE, however ACE developed along-side this market and contributed to the launch of an entire Superforecasting industry, led to other spin-off programs like OSI, CREATE, and HFC, and constituted the world’s largest forecasting experiment.
  • Little Horned Owl, a program similar to GHO, launched in 2018 and completed in 2022, sought to develop ultra-quiet mini UAVs (defined as having a take-off weight of 55 pounds or less) to further enable critical intelligence and military missions. Two different developed designs will be available for transition to government users. Each design has a flight radius of 30 miles, with 30 minutes time-on-station, while carrying a 10-pound payload.
  • IARPA utilized several innovative programs and one seedling to aid the IC and help the U.S. combat the coronavirus (SAR-COV-2). These included:
    • The Crowdsourcing Evidence, Argumentation, Thinking and Evaluation (CREATE) program. Roughly 100 Australian researchers from the country’s eight leading universities used a collaboration platform developed for the CREATE program, which launched in 2016, to analyze possible outcomes of COVID-19 policy alternatives and deliver a report to the Health Ministry and Chief Medical Officer.
    • The Functional Genomic and Computational Assessment of Threats (Fun GCAT) program. Launched in 2017, performers at Harvard University used the Fun GCAT pipeline to analyze COVID-19 genes to help reveal how COVID-19 disrupts human immune systems and what makes the virus pathogenic.
    • The Molecular Analyzer for Efficient Gas-phase Low-power Interrogation (MAEGLIN) program. A small, portable gas sensor that we originally developed through our MAEGLIN program, which launched in 2017, to identify illicit activity indicators, such as narcotics production, was used in a clinical trial we funded to determine if it can be used as a breath sensor to detect signs of Acute Respiratory Distress Syndrome (ARDS) early enough to improve patient chances of surviving COVID-19 complications. Results suggested we can distinguish COVID patients from healthy patients and monitor the progress of the disease.
    • The Finding Engineering-Linked Indicators (FELIX) program. The MIT-Broad Foundry, a performer team on the FELIX program, which launched in 2019, analyzed the publicly available SARS-CoV-2 genome using their FELIX bioinformatics pipeline in order to test the veracity of online stories claiming that SARS-CoV-2 was engineered in a laboratory. They compared the SARS-CoV-2 genome against 58 million sequences, including genomes from closely- and distantly-related viruses and in only 10 minutes, determined that all SARS-CoV-2 regions genome match naturally-occurring coronaviruses better than they match any other organisms, including any other viruses. The analysis indicated that no sequences from foreign species have been engineered into SARS-CoV-2.
    • The BioHeat seedling project at the Baylor College of Medicine provided further evidence that SARS-CoV-2 was not genetically engineered. In April 2020, Baylor developed a software pipeline to analyze protein stability and relative mutation rate. This work was aimed at faster therapeutic and vaccine discovery, and their mutation hotspot visualizations may assist with contact tracing.


Collectively, IARPA continues to focus on a range of programs that incorporate research areas such as quantum technology, computer architecture, microelectronics, data analytics, energy storage (batteries), biometrics, linguistics, site modeling, active smart textiles, radio frequency communications, and orbital debris.