About GuardAI

A team with experience in risk management and AI research

Our team brings decades of enterprise risk management experience and published AI security research to the practical problems of running LLM applications.

GuardAI gives enterprises a dedicated policy layer for AI deployments. It applies your rules to prompts and responses, independently of the model provider. Test those rules against public benchmarks and your own application traffic.

How GuardAI started.

2015

The first meeting

Jenny Yu and Professor Haohan Wang connected at Carnegie Mellon's Innovation Summit, laying the foundation for GuardAI.

June 2023

The catalyst

As the European Parliament adopted its negotiating position on the EU AI Act, Jenny wrote “You Cannot Manage AI without an AI Tool,” recognizing LLMs challenged traditional risk frameworks.

August 2023

The breakthrough

Jenny discovered Professor Wang had authored a comprehensive survey on trustworthy and aligned ML, systematically analyzing robustness, adversarial defense, interpretability, and fairness. She reached out to discuss applying the research to enterprise AI.

2023 to present

GuardAI today

Building policy enforcement for enterprise AI, drawing on research into model security and practical experience managing business risk.

Meet the founders.

Combining decades of expertise in AI risk management and trustworthy ML research.

Jenny Yu

Founder

Himalaya Quantitative Solutions (HQS)

25 years of experience in AI/ML/model risk management at major global financial institutions including FHLBank Pittsburgh, BNY Mellon, and PNC Financial Services.

Specializes in designing holistic AI/ML/model risk management frameworks and building validation/benchmarking tools to mitigate risk in a systematic and automatic way.

2013 Global Innovation Winner at PNC, led development of validation models and automated benchmarking processes that increased coverage from 5% to 99%.

Leads HQS in creating a trustworthy AI tool for Generative AI risk management.

Recognized for empowering senior leadership by transforming complex data into actionable insights, bridging gaps between intricate data, algorithms, and business processes.

Offers strategic solutions that resonate with stakeholders' needs, pragmatically and effectively.

Asian American Executive Program, Stanford University Graduate School of Business (2022). Master of Science with double majors in Quantitative Finance and Real Estate from UW-Madison.

Haohan Wang

Technology Advisor

Assistant Professor, UIUC

LinkedIn Research

Assistant Professor in the School of Information Sciences at the University of Illinois Urbana-Champaign (UIUC), with appointments across the Siebel School of Computing and Data Science, Carl R. Woese Institute for Genomic Biology, and National Center for Supercomputing Applications.

7,300+ citations on Google Scholar, with research featured at top-tier AI conferences including ICLR (Top 1% Oral) and CVPR (Top 2% Oral).

Research focuses on LLM security, adversarial robustness, and AI alignment with a decade of experience developing trustworthy machine-learning methods, from rigorous statistics to powerful deep learning and large language models.

His groundbreaking research on jailbreaking LLMs and defending against jailbreaking attacks, featured at NeurIPS 2024, directly advances the science behind AI guardrail technologies. Learn more about the research that forms the scientific backbone of GuardAI's protection mechanisms.

Authored comprehensive survey on trustworthy and aligned machine learning, systematically analyzing robustness, adversarial defense, interpretability, and fairness.

Recognized as Top 50 AI+X Rising Young Scholar by Baidu Inc. (2022).

Ph.D. in Computer Science from Carnegie Mellon University (Language Technologies Institute).

Research-driven technology.

GuardAI's protection mechanisms are grounded in peer-reviewed academic research from Professor Wang's extensive work on trustworthy machine learning. Our technology has been rigorously tested against state-of-the-art attack methodologies documented in top-tier conferences like NeurIPS 2024.

By combining Jenny's 25 years of enterprise risk management experience with Professor Wang's decade of research into AI trustworthiness, GuardAI delivers solutions that are both academically rigorous and practically deployable in production environments.

Explore our research foundation

Your AI speaks for your business.

A refund promise, a pricing detail, or a piece of advice has consequences for your business. Your assistant needs to follow the same rules as the people it works alongside.

We build GuardAI to put those rules into practice: define the boundaries, check prompts and responses, and record policy decisions for review.

Let's protect your AI deployment.

Whether you're evaluating safety options for a new product or hardening an existing system, we can show you exactly what GuardAI catches, and what your current setup misses.

Try interactive demo Contact us