Open to opportunities

Michael Schmid

@michaelschmid1

Message

Frontier AI Safety Expert with extensive experience in adversarial evaluation.

United States

Message

What I'm looking for

I seek a role that fosters innovation in AI safety and offers opportunities for impactful contributions to high-stakes AI deployment.

I am a Frontier AI Safety Expert with over 10 years of experience specializing in adversarial evaluation, red teaming, and threat modeling of frontier systems, including agentic AI and large language models (LLMs). My work has involved advising prestigious organizations such as NASA, Ford, DARPA, and the U.S. military on scalable safeguards and responsible scaling practices for high-stakes AI. I have pioneered a cutting-edge evaluation framework for misbehavior in frontier AI, which serves as a foundation for safety benchmarks and deployment-readiness evaluations.

With a PhD from MIT focused on AI safety and adversarial risk modeling, I have developed innovative evaluation frameworks and safety evaluation tools that have been integrated into CI/CD workflows. My contributions to international AI safety standards, particularly ISO 21448, have shaped the technical direction and safety evaluation practices for frontier agentic systems. I am passionate about teaching and sharing my expertise with executives and stakeholders to ensure the responsible deployment of AI technologies.

Experience

Work history, roles, and key accomplishments

Current

AI Threat Modeling & Evaluation Lead

Current

MIT

Jan 2020 - Present (5 years 6 months)

Led adversarial evaluations of agentic frontier AI systems, identifying jailbreaks, ASL-3/4-level threats, and alignment failures. Developed first-of-its-kind evaluation frameworks for high-stakes agentic AI, now informing scaling and model-readiness protocols for deployment of frontier AI systems.

Threat Modeling Jailbreak Detection Agentic AI LLMs)

Safety Evaluation Tooling Lead

German Aerospace Center (DLR)

Jan 2016 - Present (9 years 6 months)

Created safety evaluation tools deployed into automated simulation pipelines across containerized environments. Built high-throughput evaluation pipelines for auditing AI development at scale, deployed in CI/CD environments for model readiness evaluation.

Automated Simulation Docker CI CD Python

Program Lead, Frontier Autonomy

RTX

Jan 2016 - Present (9 years 6 months)

Led a $4M program and 20-person team to develop and deploy safety-critical frontier autonomy under tight, aerospace-grade safety and time constraints. Developed Python simulation tools for agentic robotics missions.

Program Management Safety Critical Autonomy Python Team Leadership

AI Safety Expert Consultant

Msg Plaut

Jan 2016 - Present (9 years 6 months)

Retained as a subject matter expert to guide strategy and deployment protocols for high-stakes AI. Contributed original language to ISO 21448, shaping technical direction and safety evaluation practices for frontier agentic systems.

Agentic AI Subject Matter Expertise

Aerospace Safety Engineer

Liebherr-Aerospace

Jan 2011 - Present (14 years 6 months)

Led root-cause analysis into high-severity in-flight spoiler deployment failures, developing reusable tooling for adversarial safety evaluation. Troubleshot fully assembled A380 aircraft under extreme time pressure during flight test and delivery phases.

Root Cause Analysis Aerospace Engineering Troubleshooting A380 Aircraft Systems

Education

Degrees, certifications, and relevant coursework

Massachusetts Institute of Technology

Ph.D., Safety & Adversarial Evaluation of High-Stakes Agentic AI

2020 - 2025

Grade: 4.9 / 5.0 GPA

Studied failures of LLM-, CNN-, KNN-, and SVM-based AI in high-stakes frontier agentic systems. Uncovered recurring risk patterns and a path enabling more effective adversarial safety evaluation.