Data Scientist, Integrity Measurement

📋 Description

• • Join OpenAI's Applied Foundations team, a critical unit focused on safeguarding our advanced AI technologies against a spectrum of adversarial threats and ensuring platform integrity as we scale.
• • As a Data Scientist, Integrity Measurement, you will be at the forefront of defending our platforms against sophisticated misuse, including financial abuse, large-scale attacks, and other malicious activities that could compromise user experience or operational stability.
• • The Integrity pillar within Applied Foundations is specifically tasked with developing and deploying scaled systems to detect and respond to malicious actors and harmful content across OpenAI's platforms.
• • This role is pivotal in maturing our systems that address severe usage harms, requiring a data scientist to meticulously measure the prevalence of these issues and evaluate the effectiveness of our countermeasures.
• • You will own the measurement strategy and quantitative analysis for a portfolio of severe, actor- and network-based usage harm verticals, delving deep into the complexities of identifying and quantifying these threats.
• • A core responsibility will be to develop and implement innovative, AI-first methodologies for prevalence measurement and other critical production safety metrics, potentially incorporating off-platform indicators and unconventional datasets.
• • You will be instrumental in building robust metrics suitable for performance goal setting and A/B testing, especially when direct prevalence metrics are not feasible or sufficient.
• • Take ownership of comprehensive dashboards and metrics reporting for your assigned harm verticals, providing clear, actionable insights to stakeholders.
• • Conduct in-depth analyses to generate insights that directly inform and drive improvements in review processes, detection algorithms, and enforcement strategies, significantly influencing product and safety roadmaps.
• • Play a key role in optimizing Large Language Model (LLM) prompts specifically for the purpose of accurate and efficient measurement of harmful activities.
• • Collaborate closely with other safety teams across OpenAI to gain a deep understanding of emerging and existing safety concerns, and to contribute to the creation of relevant policies that effectively support our safety objectives.
• • Prepare and present key metrics for leadership reviews and external reporting, ensuring transparency and accountability.
• • Develop and implement automation solutions, leveraging OpenAI's cutting-edge agentic products, to enhance efficiency and scale your own impact.
• • This role demands a proactive approach to identifying and mitigating risks, requiring a keen analytical mind and a commitment to upholding the highest standards of AI safety and integrity.
• • You will work with sensitive and potentially disturbing content, including material related to sexual violence, child safety, and other forms of severe harm, requiring emotional resilience and a strong ethical compass.
• • Contribute to the continuous evolution of our safety measurement frameworks, ensuring they remain state-of-the-art and effective against evolving threats.
• • The position offers a unique opportunity to shape the safety and integrity of globally impactful AI technologies, making a tangible difference in protecting users and the broader ecosystem.
• • You will be part of a dynamic and collaborative environment, working alongside world-class researchers and engineers dedicated to responsible AI development.
• • This role is ideal for a senior data scientist with a proven track record in trust and safety, eager to drive measurement direction and tackle some of the most challenging problems in AI safety.
• • Your work will directly contribute to OpenAI's mission of ensuring that artificial general intelligence benefits all of humanity by building trust and safety into the core of our AI systems.
• • The San Francisco or New York office location offers a vibrant work environment, with potential for urgent escalations outside of normal work hours, reflecting the critical nature of the role.
• • Embrace the challenge of measuring and mitigating complex harms in a rapidly evolving technological landscape, contributing to the responsible deployment of AI.