AI Security Engineer

Booz Allen Hamilton•Bethesda, MD

4d•$99,000 - $225,000

About The Position

As an AI Security Engineer, you will focus on identifying, measuring, and mitigating AI vulnerabilities to evasion attacks, privacy attacks, data poisoning attacks, and misuse of large language models. You will help research, develop, and implement adversarial machine learning algorithms and execute AI red teaming operations for clients across both the private and public sectors. You'll serve as the client interface and maintain responsibility across the entire AI security lifecycle, including requirements gathering and analysis, process and systems definition, data analysis, presentation of analysis to clients in a format they can digest, and development of algorithm driven products and solutions. Join us. The world can’t wait.

Requirements

5+ years of experience programming in Python
Experience with machine learning, including adversarial machine learning met hods
Knowledge of adversarial machine learning, large language models, or differential privacy
Knowledge of mathematics and statistics, including the theory of probability, statistical inference, algorithms, linear algebra, and calculus
Ability to communicate results to both technical and non-technical audiences effectively
TS/SCI clearance with a polygraph
Bachelor's degree in Computer Science, Statistics, Mathematics, Physics, Applied Mathematics, or Engineering

Nice To Haves

Experience with application areas of machine learning, including computer vision, large language models, natural language processing, and learning on graphs
Experience with Bayesian deep learning and Gaussian processes
Experience building complex data pipelines
Experience using GPUs for machine learning using frameworks, including PyTorch or TensorFlow
Knowledge of cloud systems, including AWS, Azure, or GCP
Ability to work independently on complex tasks

Responsibilities

identifying, measuring, and mitigating AI vulnerabilities to evasion attacks, privacy attacks, data poisoning attacks, and misuse of large language models
research, develop, and implement adversarial machine learning algorithms
execute AI red teaming operations for clients across both the private and public sectors
serve as the client interface and maintain responsibility across the entire AI security lifecycle, including requirements gathering and analysis, process and systems definition, data analysis, presentation of analysis to clients in a format they can digest, and development of algorithm driven products and solutions