Usman Gohar

Hi! I am Usman! I am a final-year PhD candidate and F. Wendell Scholar in Computer Science at Iowa State University, advised by Dr. Robyn Lutz in the Laboratory for Software Safety. My research interests lie at the intersection of AI Safety and Software Engineering, with a focus on evaluating and mitigating algorithmic harms and improving responsible AI practices.

Beyond academic research, I am also very interested in AI Safety Policy and Governance. I served as the Section Lead for “AI-generated content and criminal activity” for the 2026 International AI Safety Report, a global initiative chaired by Turing Award winner Yoshua Bengio. Led by the UK AISI, the report brings together over 100 leading experts, 36+ governments, and organizations to assess the frontier risks posed by advanced AI systems.

I am also one of the founding members and chairs at the Evaluating Evaluations Coalition (EvalEval), an initiative that brings together researchers and practitioners to examine the reliability, validity, and governance of AI evaluation methodologies and benchmarks.

My research has focused on creating robust, practical approaches to operationalize AI and software safety within complex, data-driven systems, including advancing fairness, mitigating harms from machine learning and AI, and improving safety assurance for autonomous systems (e.g., drones) and AI systems. I’ve published peer-reviewed research in leading venues in software engineering, machine learning, and AI ethics.

My research focuses on three key areas:

• Operationalizing AI and Software Safety Developing practical frameworks and methodologies to apply system safety principles in complex, data-driven AI systems. My work includes advancing safety assurance approaches for autonomous drones and evaluating risks associated with generative AI systems.

• Algorithmic Fairness and Harm Mitigation
  Investigating how harms from machine learning models manifest and how fairness can be systematically measured and improved. I analyze existing fairness metrics and propose improvements to reduce bias and unfairness.

• Evaluation and Deployment of Safe AI
  Designing transparent evaluation techniques that bridge theoretical AI safety concepts with real-world deployment challenges, focusing on creating actionable tools and guidelines for reliable, fair, and safe AI systems.

Previously, I have worked as a Data Scientist in different sectors like agriculture, manufacturing, and power systems, specializing in forecasting, predictive analytics, and model deployment. I am passionate about translating foundational AI research into practical applications.

View my CV here