About AIUC

Our mission is to underwrite superintelligence. We red-team, certify, and insure agents, e.g. ElevenLabs, UiPath, Intercom. Soon, agents will take on critical work in hospitals, banks and governments. This could create abundance - or go terribly wrong. History teaches us that standards and insurance are capitalism’s best tools for making risk legible and shaping incentives. We’ve built the gold standard for agent security & reliability, AIUC-1, with 100+ CISOs from the Fortune 1000. Our team comes from Anthropic, McKinsey, METR, Weights & Biases, Perplexity, Thiel Fellowship, etc. Nat Friedman led our $15m seed. Read more here, here, and here.

The role

You’ll lead the red-teaming of the world’s most used AI agents, working directly with our customers' agents end-to-end from scoping to certification.

Refine the core engine that runs all our red-teaming evals
DRI end-to-end: technical scoping, reconnaissance, execution, remediation, certification
Productize frontier adversarial attacks
Taste-test the evals to set the quality bar
Productize your workflow with agents
Figure out how we eval new types of agents, e.g. computer use

You thrive across tinkering with complex AI workflows and translating technical trade-off to CISOs.

Strong Back - High standards, keep our promises, work hard

Fast Feet - Experiment, cut scope, take delight in pace

Eyes Up - ‘Feel the AGI’ across our strategy, teams and day-to-day work

Open Heart - Say the scary thing, create connection, make room for emotions