Description
Job Profile
Help secure AI products and models that are introducing industry-novel security risks and pushing established security boundaries
Conduct secure design reviews and threat modeling for product teams — identifying risks, attack surfaces, and prioritizing mitigations early.
Champion shift-left security practices — building security into the SDLC from ideation through deployment.
Partner with product, and engineering teams to integrate security controls into distributed systems — including Micro-service architecture, cloud services, container platforms, and APIs.
Support penetration tests, red team findings, and bug bounty remediation Surface security considerations for new domains such as AI/ML features, model pipelines, and agent frameworks.
Required
The ability to think like an attacker, develop threat models and help teams reason through different approaches to reducing security risk
Experience securing large-scale, distributed, cloud-native systems
Experience in application/product/security engineering or a related technical security role
Experience securing multi cloud platforms (AWS, GCP, Azure), multi-tenant SaaS, containerization and orchestration technologies such as Docker or Kubernetes
Hands-on experience with secure development tools (SAST/DAST/IaC scanning), secure code review, and penetration testing
Familiarity with at least one programming language (e.g., Java, Python, Go, JavaScript)
Excellent communication skills — able to translate security risks to technical and non-technical stakeholders.
Preferred / Nice to have
Familiarity with AI/ML security risks (prompt abuse, model poisoning, RAG data leakage).
Experience with or leading bug bounty, red team, or penetration testing programs.
Contributions to open-source security projects or tools.
For roles in San Francisco and Los Angeles: Pursuant to the San Francisco Fair Chance Ordinance and the Los Angeles Fair Chance Initiative for Hiring, Salesforce will consider for employment qualified applicants with arrest and conviction records.