About Sycamore

Sycamore is building the trusted agent operating system for the enterprise. Our platform helps companies build, deploy, and orchestrate agentic apps that take on real operational work, with the security and control large organizations need.

We are a small, engineering-led team working directly with Fortune 500 enterprises. We have raised $65M from Coatue and Lightspeed, along with other investors and industry leaders.

The role

You will secure the product and platform that run Sycamore's enterprise agents. The surface spans user-facing applications, APIs, agent runtimes, tools and connectors, identity and authorization, untrusted code execution, multi-tenant data, cloud infrastructure, and the software-delivery system itself.

This is a hands-on engineering role. You will write production code, build security controls and paved paths, review designs and code, model threats, test systems from an attacker's perspective, investigate incidents, and work with engineers until durable fixes reach production.

Agent systems introduce unusual security problems. A model may consume hostile instructions from a document, invoke tools on a user's behalf, generate executable code or browser content, retain information in memory, and operate across long-running workflows. Securing that system requires more than a conventional web checklist: identity must remain attributable, authority must narrow at every boundary, untrusted content must stay contained, credentials must remain outside model context, and high-impact actions must be observable and governable.

You will work across Product, Core AI, and Infrastructure. Those teams continue to own their systems; Security helps them identify risks early, provides reusable controls, validates important boundaries, and steps in directly when a problem needs deep security engineering.

What you will do

Lead threat modeling and security architecture for agent products, APIs, connectors, execution environments, customer applications, and platform services.

Build production security controls for authentication, authorization, delegated agent identity, workload identity, credential brokering, tenant isolation, encryption, policy enforcement, and auditability.

Design isolation boundaries for untrusted code and generated content, including sandboxing, network and resource controls, browser containment, and safe communication with trusted services.

Identify and remediate vulnerabilities through design review, code review, targeted security assessment, hands-on exploitation, and adversarial testing.

Develop reusable libraries, services, policies, test harnesses, and developer tooling that eliminate vulnerability classes rather than documenting the same mistake repeatedly.

Secure the software-development lifecycle through dependency verification, secret detection, code and infrastructure scanning, build isolation, artifact integrity, and risk-based release gates.

Build security telemetry, detections, investigation tools, containment paths, and response automation across application, identity, agent, and infrastructure events.

Participate in incident response from triage and containment through root-cause analysis, remediation, evidence preservation, and control improvement.

Establish practical vulnerability-management practices with clear severity, exploitability, ownership, remediation, verification, and exception handling.

Partner with Product and Core AI on risks specific to agentic systems, including prompt injection, confused deputies, excessive agency, unsafe tool calls, memory poisoning, sensitive-data leakage, and model-generated executable content.

Partner with Infrastructure on cloud and cluster security, network boundaries, workload identity, secrets, data protection, deployment safety, and customer-controlled environments.

Translate enterprise security requirements into defensible engineering controls and evidence without allowing questionnaire-driven work to replace actual risk reduction.

Raise the security capability of the engineering organization through clear design patterns, direct coaching, useful reviews, incident learning, and tools engineers want to use.