GEORGIA TECH RESEARCH CORP

• TRIAD NATIONAL SECURITY, LLC

  UEI X7WUS5LRBQU3

  $811.8K

  Subcontractor to this prime · 2 sub-awards · latest Aug 6, 2024

  • Department of Energy
  • “GEORGIA INSTITUTE OF TECHNOLOGY-STANDALONE-ITEM LEVEL-R AND D”
• UNIVERSITY OF FLORIDA

  UEI NNFQH1JAPEP3 · not in the directory yet

  $670.2K

  Subcontractor to this prime · 3 sub-awards · latest Oct 2, 2025

  • Department of the Air Force
  • “D WHILE ALSO ENSURING THAT ASSETS STILL PERFORM AT A USEFUL LEVEL. WE WILL QUANTIFY TRADEOFFS IN THIS CONTEXT BY COMPARING THE COST INCURRED WITH PRIVACY TO THE COST INCURRED WITHOUT PRIVACY. WE WILL COMPUTE THIS DIFFERENCE IN TERMS OF THE STRENGTH OF PRIVACY PROTECTIONS, AND FROM THIS WE WILL DEVELOP GUIDELINES FOR THE CALIBRATION OF PRIVACY. THRUST 2 THE CONTROLLER USED BY AN ASSET CAN ALSO BE SENSITIVE BECAUSE IT ENCODES HOW AN ASSET MAKES DECISIONS, AND WE WILL DEVELOP DECEPTIVE AUTONOMY TECHNIQUES TO PROTECT IT. WITHOUT ANY PROTECTIONS, A CONTROLLER COULD BE UNCOVERED BY AN ADVERSARY THAT OBSERVES AN ASSET, E.G., THROUGH USING MACHINE LEARNING TECHNIQUES, AND THIS CAN ENABLE AN ADVERSARY TO PREDICT AN ASSET’S FUTURE DECISIONS. MATHEMATICALLY, A CONTROLLER IS A FUNCTION THAT MAPS THE STATE OF AN ASSET TO A CONTROL INPUT. THE PRIVATIZATION OF SUCH A MAPPING WILL THEREFORE BE DONE BY MAKING IT APPROXIMATELY INDISTINGUISHABLE FROM OTHER FUNCTIONS THAT MAP BETWEEN THE SAME SPACES. THIS WILL BE DONE BY ADDING A RANDOM PERTURBATION TO IT. OF COURSE, WE WISH TO PRESERVE CERTAIN USEFUL PROPERTIES OF FEEDBACK CONTROLLERS SUCH AS CONTINUITY OR SMOOTHNESS, AND WE WILL DO SO BY POST-PROCESSING THE PERTURBED CONTROLLER. BECAUSE DIFFERENTIAL PRIVACY IS NOT HARMED BY POST-PROCESSING, THESE OPERATIONS CAN BE FREELY APPLIED, AND WE WILL CO-DESIGN A PRIVACY MECHANISM (FOR PERTURBING THE CONTROLLER) AND POSTPROCESSING TECHNIQUES (FOR IMPROVING ITS MATHEMATICAL PROPERTIES) TO SIMULTANEOUSLY PROVIDE PRIVACY AND ENSURE THE UTILITY OF PRIVATIZED CONTROLLERS. NEXT, WE WILL QUANTIFY TRADEOFFS BETWEEN PRIVACY AND PERFORMANCE BY QUANTIFYING THE COST INCURRED BY A CONTROLLER AND COMPARING THAT TO THE COST INCURRED BY ITS PRIVATIZED FORM. WE WILL BOUND THIS DIFFERENCE IN TERMS OF THE STRENGTH OF PRIVACY TO MAKE THESE TRADEOFFS EASILY TUNABLE. 1 THRUST 3 TO ENSURE THE PRACTICAL APPLICABILITY OF ALL DEVELOPMENTS, WE WILL CONDUCT EXPERIMENTS ON TEAMS OF ROBOTS THROUGHOUT THIS PROJECT. OUR EXP”
  • “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’S FUTURE DECISIONS. MATHEMATICALLY, A CONTROLLER IS A FUNCTIO”
  • “THE GEORGIA INSTITUTE OF TECHNOLOGY WILL CONTRIBUTE TO RESEARCH ON MULTI-AGENT LOCALIZATION. IN PARTICULAR, OUR WORK WILL FOCUS ON DEVELOPING NOVEL DECENTRALIZED AND DISTRIBUTED OPTIMIZATION ALGORITHMS FOR SOLVING THE POSE GRAPH OPTIMIZATION PROBLEM TO ENABLE A GROUP OF AGENTS TO LOCALIZE ITSELF IN A GPS-DENIED ENVIRON-MENT. THIS WORK WILL BUILD ON EXISTING CAPABILITIES AT THE AVL AND WILL FURTHER IMPROVE ON THE BANDWIDTH REQUIREMENTS CURRENTLY NEEDED FOR SUCCESSFUL LOCALIZATION. AS PART OF THIS WORK, A GENERALIZED OPTIMIZER WILL ALSO BE DEVELOPED FOR SOLVING COMBINATORIAL OPTIMIZATION PROBLEMS LIKE THE WEAPON TARGET ASSIGNMENT PROB-LEM IN A DISTRIBUTED, ASYNCHRONOUS WAY THAT ENABLES AGENTS TO DYNAMICALLY REACT TO A COMPLEX ADVERSARIAL ENVIRONMENT. THIS PROJECT WILL ALSO CONSIDER COMMUNICATION PROTOCOLS THAT ALLOW AGENTS TO DECIDE THE RIGHT TIME AND DATA TO COMMUNICATE TO SOLVE TASKING PROBLEM COOPERATIVELY.”
• NOU SYSTEMS INC

  UEI CSPJJ1VXPMT9

  $420K

  Subcontractor to this prime · 1 sub-award · latest Jul 25, 2024

  • Missile Defense Agency
  • “GEORGIA TECH RESEARCH CORPORATION WILL PROVIDE NSI WITH TECHNICAL SUPPORT.”
• FLUOR MARINE PROPULSION, LLC

  UEI CWHMVCX7K1N6

  $387.4K

  Subcontractor to this prime · 1 sub-award · latest Jun 16, 2025

  • Department of Energy
  • “RICKOVER FELLOWSHIP”

• REGENTS OF THE UNIVERSITY OF CALIFORNIA, THE

  UEI GS3YEVSS12N6

  $1.2M

  Subcontracted by this company · 1 sub-award · latest Jun 21, 2024

  • Defense Advanced Research Projects Agency
  • “DEMONSTRATE LOW NOISE FERRO-FINFET OPERATING WITH <1FJ SWITCHING ENERGY, 5NS READ SPEED, 1010 READ ENDURANCE. DESIGN AND DEMONSTRATE FERRO-MCE DEVICE WITH CELL SIZE OF 6F2. DEVELOP COMPACT MODELS OF IN-HOUSE FERRO-FINFETS AND FOUNDRY-GRADE FERRO-MOSFETS THAT ARE CAPABLE OF SUPPORTING CYCLE-ACCURATE MAM ARRAY AND CIMA PROTOTYPE DESIGN AND SIMULATION.”
• OREGON STATE UNIVERSITY

  UEI MZ4DYXE1SL98 · not in the directory yet

  $250K

  Subcontracted by this company · 1 sub-award · latest Dec 11, 2025

  • Defense Advanced Research Projects Agency
  • “OREGON STATE UNIVERSITY (SUBCONTRACT PI DR. ZANE MA) WILL JOIN THE PROGRAM FOR PHASE 2 AND PERFORM THE FOLLOWING TASKS FROM M34 – M46: 1) DESIGN, BUILD, AND EVALUATE A NEW TASK 2.5.4.1 – CASE STUDY: TESTBED + WEB3 ALONG WITH SUBCONTRACT STONY BROOK UNIVERSITY (SUBCONTRACT PI DR. NICK NIKIFORAKIS) AND 2) COORDINATE REPORTING AND EVALUATION OF THE TASK TO GT LEAD AND DARPA. DETAILED RESPONSIBILITIES FOR THE NEW TASK ARE SPECIFIED BELOW. TASK NUMBERING AND TIMELINE IS CONSISTENT WITH THE FULL PROJECT SOW (M46 = SEPTEMBER 2026).”
• ANSATZ AI LLC

  UEI C3JRD29U38L1

  $100K

  Subcontracted by this company · 1 sub-award · latest Sep 14, 2026

  • Department of the Army
  • “PREDICTING MATERIAL PERFORMANCE OF CROSS-LAMINATED TIMBER (CLT) UNDER HIGH-VELOCITY LOADING IS AN UNMET CHALLENGE. MACHINE LEARNING EXCELS AT PREDICTIVE MODELING OF THE RESPONSES OF COMPLEX SYSTEMS BUT REQUIRES LARGE DATASETS OF PRIMARY DATA TO DEVELOP AN ACCURATE RESPONSE SURFACE THAT RELATES THE INPUT VARIABLES AND THE OUTPUT RESPONSES. HIERARCHICAL MACHINE LEARNING (HML) WAS DEVELOPED AT CARNEGIE MELLON UNIVERSITY (CMU) TO MODEL COMPLEX MATERIAL SYSTEMS FROM SMALL DATASETS. THE ALGORITHM RELIES ON INTEGRATION OF DOMAIN KNOWLEDGE IN THE FORM OF EMPIRICAL MODELS AND ADDITIONAL EXPERIMENTAL RESULTS TO MINIMIZE THE DATA REQUIREMENTS COMPARED TO TRADITIONAL MACHINE LEARNING. ANSATZ AI WILL WORK WITH DR. LAUREN STEWART AND HER COLLEAGUES AT GEORGIA TECH TO BUILD PREDICTIVE MODELS OF CLT PERFORMANCE AS A FUNCTION OF FEEDSTOCK AND LAYERING CHARACTERISTICS.”
• MASSACHUSETTS INSTITUTE OF TECHNOLOGY

  UEI SN5KKUDR9LY8

  $56.7K

  Subcontracted by this company · 1 sub-award · latest Apr 29, 2026

  • Defense Advanced Research Projects Agency
  • “MIT WILL PERFORM FUNDAMENTAL RESEARCH ON THE PROJECT. THE FREEFORM OPTICS DESIGN METHODOLOGY AND EXPERIMENTAL IMPLEMENTATION HAVE BEEN DISCLOSED IN PEER-REVIEWED PUBLICATIONS: PHOTON. RES. 12, 1055-1066 (2024); LASER PHOTONICS REV. 17, 2200025 (2023). THE HIGH-INDEX GLASS MATERIAL PREPARATION AND NANOIMPRINT FABRICATION TECHNOLOGY INVOLVED FOR THE MICRO-OPTICS FABRICATION HAVE ALSO BEEN PUBLICLY DISCLOSED IN: ADV. OPT. MATER. 2, 478-486 (2014); ADV. OPT. MATER. 2, 759-764 (2014). THEREFORE, THE MIT TASKS ARE FUNDAMENTAL RESEARCH IN NATURE. SOW FOR GEORGIA TECH MIT WILL PERFORM THE FOLLOWING TASKS DURING PHASE I OF THE PROJECT (04/2025 – 09/2026): 1. DESIGN OF THE MICRO-OPTICAL COUPLERS FOR EFFICIENT WAVEGUIDE COUPLING; 2. MICRO-OPTICAL COUPLER FABRICATION USING IN-HOUSE PROTOTYPING FACILITIES; 3. OPTICAL CHARACTERIZATION OF THE MICRO-OPTICAL COUPLERS; 4. PACKAGING AND ASSEMBLY OF THE MICRO-OPTICAL COUPLERS WITH PHOTONIC WAVEGUIDES. MIT WILL PERFORM THE FOLLOWING TASKS DURING PHASE II OF THE PROJECT (10/2026 – 03/2028): 1. DESIGN OPTIMIZATION OF THE MICRO-OPTICAL COUPLERS FOR EFFICIENT WAVEGUIDE COUPLING BASED ON PHASE I OUTCOME; 2. DEMONSTRATION OF SCALABLE MICRO-OPTICAL COUPLER INTEGRATION WITH PICS FABRICATED VIA STANDARD PHOTONIC FOUNDRY SERVICE; 3. HIGH-THROUGHPUT OPTICAL CHARACTERIZATION OF LARGE MICRO-OPTICAL COUPLER ARRAYS; 4. PACKAGING AND ASSEMBLY OF THE MICRO-OPTICAL COUPLERS WITH PHOTONIC WAVEGUIDE CIRCUITS.”

Agencies generally may not award to excluded parties, and primes may not subcontract with them above the threshold in FAR 52.209-6 (except COTS items) without a compelling reason reported to the contracting officer (FAR 9.405, 9.405-2). Some exclusion types apply only to specific programs: read the record on SAM.gov.

No confirmed match on the Consolidated Screening List (screened Nov 13, 2026).

The list combines Commerce, State and Treasury export-control and sanctions lists. A “no match” result is a name-and-address screen, not an export-compliance determination.