Staff+ Software Engineer, Account Compromise
London, UK
Posted 7d ago
About the job
Anthropic's Safeguards organization is responsible for building systems that ensure Claude is safe to use at scale. The Account Compromise team focuses on protecting users from losing account control and preventing compromised accounts from being used to abuse the platform. This role involves owning the architecture for detecting, containing, and remediating account compromise across Claude and its Developer Platform, making key design decisions for other engineers. The work requires deep technical expertise, adversarial problem-solving, threat modeling, and the ability to scope and lead complex, multi-month projects from ambiguous beginnings. You will operate with significant autonomy, driving alignment with various teams and shaping Anthropic's global approach to account security.
Responsibilities
- Set technical direction and own the architecture for account compromise detection, response, and remediation.
- Scope and lead complex, multi-month engineering projects from ambiguous starting points to production.
- Build and evolve detection systems for account takeover, credential abuse, and compromised API keys in near real-time.
- Design automated response flows to mitigate attacker access while minimizing user disruption.
- Lead end-to-end investigations of significant compromise incidents and translate learnings into automated defenses.
- Threat model attacker adaptations and prioritize team work based on future threats.
- Drive cross-organizational alignment on account security direction with Security, Product, Support, and Policy teams.
- Define and measure team success metrics.
- Set technical standards for the domain and mentor other engineers.
- Surface patterns from compromise cases to improve upstream protections.
Requirements
- 10+ years of experience designing, building, and operating detection, anti-fraud, anti-abuse, or security systems in production.
- Proven track record of independently scoping and delivering complex, ambiguous, multi-month technical projects.
- Experience making architectural decisions in adversarial domains.
- Proficiency in Python and SQL, with strong software engineering fundamentals.
- Experience leading investigations into account-based abuse or security incidents and translating findings into automated detection.
- Ability to rigorously analyze large behavioral or telemetry datasets.
- Strong written communication skills and experience driving alignment across multiple teams.
- Sound judgment regarding the trade-off between security and user disruption, with the ability to articulate and defend decisions.