- Secure how Life360 accesses frontier models. Design, build, and iterate the access controls, policy enforcement, and authorization patterns that govern how systems interact with the frontier models they rely on.
- Build secure patterns for MCP access and tool use authorization. Build and own the controls that vets, risk-tier, and govern how we integrate with external tools and services via MCP as adoption expands across engineering teams.
- Design and build the identity and authorization model for autonomous agents: service identities, scoped credentials, and least-privilege access patterns.
- Design and build agentic observability and adversarial defenses. Build the telemetry pipelines and behavioral monitoring that provide visibility into AI system behavior. Implement architecture-level defenses against prompt injection and related adversarial attack classes.
- Shape security for the common AI end-user platform. Lead design reviews, build access controls, data boundary enforcement, and abuse detection that keep a shared AI environment safe across users with different privilege levels.
- Secure the shared knowledge layer. Define access control and data governance for retrieval augmented and reasoning systems, ensuring AI-powered tools don't surface sensitive data to the wrong systems or users.
- Build AI supply chain integrity into the platform. Develop model provenance practices, service vetting, and dependency controls that keep the AI stack trustworthy as it grows.
- Partner with Privacy, Legal, and Data Platform to ensure the right controls are built into pipelines handling real-time location, family relationship data, and data involving minors.
OAuthLLM