Backed by Sequoia.
About the role
Large companies run on mission-critical internal systems that are often brittle, fragmented, and difficult to expose safely to modern AI. Security teams say the main obstacle is not detecting risk. It is getting internal databases, APIs, and legacy systems in front of an LLM stack without a months-long integration cycle. This role is to build secure AI agents that operate behind the corporate firewall while respecting existing permissions, controls, and operational boundaries.
What they're looking for
- Own the build across agent architecture, connectors, orchestration, access control, auditability, and deployment
- Design for constrained network environments where cloud-first assumptions, latency, and data movement require deliberate tradeoffs
- Build the natural-language task layer and connectors needed for a focused enterprise workflow
- Implement security primitives such as credential isolation, least-privilege execution, action sandboxing, approval gates, and rollback
- Work with AI Fund's build team and enterprise users to validate the first wedge against real internal-tooling pain
More about this role
Large companies run on mission-critical internal systems that are often brittle, fragmented, and difficult to expose safely to modern AI. Security teams say the main obstacle is not detecting risk. It is getting internal databases, APIs, and legacy systems in front of an LLM stack without a months-long integration cycle. This role is to build secure AI agents that operate behind the corporate firewall while respecting existing permissions, controls, and operational boundaries.
- An agent framework that connects securely to internal databases, APIs, file systems, and legacy interfaces behind corporate firewalls.
- A task-orchestration layer that turns operator intent into safe, reviewable, multi-step actions across internal systems.
- An authentication and access-control layer that integrates with enterprise identity systems and preserves existing permission boundaries.
- A monitoring and audit system that records every agent action for compliance, debugging, approval, and rollback.
- Own the build across agent architecture, connectors, orchestration, access control, auditability, and deployment.
- Design for constrained network environments where cloud-first assumptions, latency,...
Browse similar: Startup jobs · San Francisco Bay Area