Software Engineer, Ads Integrity

Responsibilities

• Design and build real-time and offline systems that detect harmful, deceptive, fraudulent, or policy-violating ads, advertisers, creatives, and landing experiences.
• Develop scalable risk-signal pipelines, rules and model-serving infrastructure, decision systems, and enforcement workflows across the ads lifecycle.
• Build advertiser-verification, account-risk, abuse-prevention, and fraud-detection capabilities that raise the cost of adversarial behavior.
• Create review and investigation tools that help operations and policy teams make fast, consistent, and explainable decisions.
• Design feedback loops, labeling systems, and integrity metrics that improve detection quality while managing false positives and user impact.
• Partner with Ads Delivery and Ads ML to integrate integrity checks into retrieval, ranking, auctions, pacing, and serving without compromising reliability or latency.
• Engineer resilient systems that adapt to changing adversarial tactics and support safe experimentation and effective incident response.
• Define the technical strategy and roadmap for ads integrity across OpenAI’s monetization stack.
• Operate systems with high engineering rigor through testing, observability, auditability, privacy and security reviews, and strong operational practices.

Requirements

• 10+ years of experience building and operating large-scale distributed systems, ideally in ads integrity, trust and safety, anti-abuse, fraud, security, risk, or an adjacent domain.
• Understand adversarial systems and have built detection, decisioning, enforcement, verification, or investigation capabilities.
• Have combined rules, heuristics, machine-learning signals, human review, and feedback loops into dependable production systems.
• Can reason about precision and recall, false positives, explainability, auditability, appeals, and the operational effects of automated enforcement.
• Have built low-latency, high-throughput services or data pipelines with demanding reliability, correctness, privacy, and security requirements.
• Comfortable partnering with policy, operations, legal, safety, security, product, and machine-learning teams on ambiguous and sensitive problems.
• Think holistically across threat models, architecture, data quality, observability, user impact, and advertiser trust.
• Can define technical direction in an ambiguous 0→1 environment and independently drive complex work across teams.
• Communicate clearly and make technical decisions grounded in safety, fairness, system health, and long-term ecosystem quality.