AI Model Evaluation and Trustworthy AI Testing

This subcontract, titled AI Model Evaluation and Trustworthy AI Testing, involves providing specialized testing and evaluation services for Large Language Models supporting prime contractors on Department of Veterans Affairs projects. The scope of work focuses on ensuring model integrity through bias evaluations using prompt pairs, measuring benchmark scoring for accuracy and honesty, and conducting Red Teaming exercises to mitigate output bias. All activities must align with Trustworthy AI principles and utilize dedicated AI system evaluation tools to compare model outputs. Managed by the Technology Acquisition Center Nj under NAICS code 541511, the contract requires the delivery of comprehensive AI System Development and Operation Documentation along with detailed Bias Evaluation reports. The opportunity was posted on October 30, 2026, with a response deadline of November 6, 2026.