Careers
Careers at Optimus Labs
We are building the security layer for AI agent adoption in the enterprise.
That means discovering, mapping, inspecting, and protecting the SKILLs, MCP servers, plugins, and agent harnesses people run with their own access, across endpoints and cloud environments. Optimus Labs is early-stage, so every person who joins changes what the company can take on.
Agentic AI security / Research-led / Early-stage
Why this work matters
Four problems the team is working on right now.
- 01
A new attack surface is already inside
SKILLs, MCP servers, plugins, and agent harnesses install in minutes and run with a person's own access. EDR, AppSec, and DLP read that activity as innocuous user behavior.
- 02
Start from what is actually there
Most enterprises cannot list the agent assets their people are running. We discover each one, tie it to the person it runs as, and map the data, credentials, and outbound actions it can reach.
- 03
Posture with evidence, not vibes
Static analysis of harness configuration, permissions, tools, and prompts, mapped to the OWASP Top 10 for LLM Applications, with findings a security team can act on.
- 04
Protection while the agent runs
Runtime monitoring of agent activity with alert and block options, so a single run never combines untrusted input, sensitive data, and external action.
Open roles
No public openings right now
We post every role on this page before we post it anywhere else. If you are already doing this work, send an introduction and we will write when a relevant role opens.
How we work
Research first, then product
Before Optimus Labs was a product, it was research. We investigate real attacks on AI agent assets and publish the evidence, including the open-source scanners that came out of those investigations.
Read the Civilizations briefings
