Senior Product Security Engineer

Adobe Adobe · Enterprise · New York, NY +1

Senior Product Security Engineer at Adobe responsible for designing and delivering AI-driven security features, integrating Azure OpenAI, RAG, and managing AI risks like prompt injection. The role involves shaping platform architecture, defining evaluation methods for LLM outputs, and guiding other engineers. Requires strong software development background, secure SDLC practices, and hands-on experience with production AI systems.

What you'd actually do

  1. Own the design and delivery of AI-driven security features, from Azure OpenAI integrations to retrieval-augmented generation and context retrieval.
  2. Shape the architecture of the team's platforms across React, Python FastAPI, Celery, Postgres, Redis, and Kubernetes.
  3. Define how the team evaluates LLM and retrieval outputs so security analyses stay accurate and reliable.
  4. Lead the team's approach to AI risks like prompt injection, data exposure, and output manipulation.
  5. Spot gaps in security coverage and build capabilities that deepen findings across products.

Skills

Required

  • Secure SDLC practices
  • application security
  • threat modeling
  • Python
  • JavaScript
  • React
  • production AI systems with LLMs
  • prompt engineering
  • AI APIs
  • retrieval-based systems
  • AI risks like prompt injection
  • hallucination
  • evaluating outputs for quality and safety
  • system architecture design
  • cloud platforms (Azure preferred)
  • containers
  • CI/CD pipelines
  • Git
  • modern workflows
  • aligning teams
  • leading technical work
  • explaining risk to technical and non-technical audiences
  • mentoring engineers
  • lifting security practices across a team

Nice to have

  • React preferred
  • Azure preferred

What the JD emphasized

  • production AI systems
  • AI risks like prompt injection
  • evaluating outputs for quality and safety

Other signals

  • AI-driven security features
  • Azure OpenAI integrations
  • retrieval-augmented generation
  • LLM and retrieval outputs evaluation
  • AI risks like prompt injection
  • production AI systems with LLMs