Staff AI Safety Engineer — Agent Platform & Risk

Harbor Behavioral Health; San Francisco (CA); Sur place; USD 190 000 - 270 000; Plein temps

Avantages offerts par ce poste

Early-stage equity

Medical, dental & vision for you and 3

Flexible paid time off

Hardware budget

Relocation to San Francisco

Résumé du poste

Harbor is hiring Members of Technical Staff to build safe, insurable AI agents across research, product, and deployment teams in San Francisco. You will work on end-to-end systems—from evaluation harnesses to runtime controls—that ensure agents operate within regulatory and client guidelines.

You’ll collaborate across disciplines to ship features quickly, with a founder mindset and a focus on practical risk mitigation and reliability.

Qualifications

• Built real systems that real people depended on.
• Strong in at least one: ML/LLM, distributed systems, security, or product engineering.
• Shipped software to production and carried a pager.
• Comfortable in a codebase defined by a conversation with a customer and a launch date.
• You want to work on agents from the angle of insurability.

Responsabilités

• Agent discovery across clouds and shadow deployments.
• Measurement: evaluations, red-teaming and risk scoring against our framework.
• Mitigation: runtime controls that stop an agent breaking a rule at machine speed.
• The platform itself: product, SDKs, integrations, and data connectivity.
• Forward deployment: work with banking/insurance customers to put their agents on Harbor.

Connaissances

Distributed systems

ML/LLM systems

Security

Product engineering

Production deployments

Description du poste

Harbor is hiring Members of Technical Staff to build safe, insurable AI agents across research, product, and deployment teams in San Francisco. You will work on end-to-end systems—from evaluation harnesses to runtime controls—that ensure agents operate within regulatory and client guidelines.

You’ll collaborate across disciplines to ship features quickly, with a founder mindset and a focus on practical risk mitigation and reliability.