AI Safety Engineer - Boston [80881]

We’re seeking a new team member to support a leading industry company's innovative efforts in AI safety and model alignment. This role offers a hybrid work arrangement—primarily remote with the option for onsite presence—and is contracted for a duration of 4 to 6 months, providing an excellent opportunity to contribute to critical safety initiatives within a dynamic environment. • • HYBRID IN OFFICE IN NEW YORK CITY OR BOSTON • • The role acts like a safety engineer + researcher whose job is to pressure-test AI systems and help design them so users get good, safe recommendations. • • Proficiency in Python, Java, and SQL, with practical experience working hands-on with complex systems and data. Responsibilities: • Collaborate with multidisciplinary teams to conduct adversarial testing and red-teaming campaigns, identifying vulnerabilities and generating training data. • Develop and oversee processes such as prompt engineering, preference tuning, and automatic prompt optimization to enhance system robustness. • Work intensively with data querying, automation frameworks, and model evaluation tools to ensure safety standards are met. • Align AI systems with evolving safety policies through scalable, robust work, including high-quality data production and system fine-tuning. • Participate in benchmark assessments, comparing system performance with comparable services to identify areas for improvement. • Study how model behavior changes with different prompts, context windows, and system instructions. • Research preference tuning strategies (e.g., reward models, human feedback, synthetic feedback). • Experiment with automatic prompt optimization to improve safety at scale. • Evaluate tradeoffs between safety, usefulness, and personalization quality. Qualifications: • Demonstrated experience in safety-related research, adversarial testing, or system alignment projects. • Proficiency in Python, Java, and SQL, with practical experience working hands-on with complex systems and data. • Extensive knowledge of large language models, prompt engineering, and safety techniques. • Pursuing or holding an MSc or PhD in an AI/ML-related field, with a focus on safety or agentic systems preferred. • Experience with cross-language models is an advantage. • Expertise in safety research, red teaming, and advanced model alignment methods. Perks and Benefits: • Medical, dental, and vision insurance • Life insurance coverage • 401(k) retirement program • Commuter benefits • eLearning and ongoing training opportunities • Education reimbursement Eligibility: • Must work 30+ hours per week • Assignment duration of at least 10 weeks If you meet the qualifications and are excited about this opportunity, apply today! Our team will connect with you to discuss next steps, support you through the interview process, and advocate for your success.