Careers

Join the watch.Build what keeps agents safe.

We build the AI defense layer that lets agents work without fear. If you care about precision and stay calm under pressure, there's a seat on the watch for you.

07Open roles
How we work

Our values are the job description.

We hire for three things before any skill on a résumé. You will see them in every review, incident and launch.

VAL.01

Vigilance

Always watching, never anxious.

You notice the small thing early. You would rather flag a risk on Tuesday than explain an incident on Friday, and you say it calmly.

VAL.02

Precision

Every pixel, packet and word is deliberate.

You measure before you claim, write it down before you ship, and treat accuracy as part of the craft rather than a final check.

VAL.03

Resilience

Systems that bend and recover.

You design for failure, recover fast, and run blameless reviews. You hold up in the incident room and in the boardroom.

Open roles

Find your post.

Showing 7 of 7

Own the engine that decides, in single-digit milliseconds, whether an agent's tool call runs, waits or stops.

  • Design the policy language teams write and review in their repos
  • Keep p50 decision latency under 6 ms as rule sets grow
  • Build the replay tooling that proves a policy change is safe

Make RAKSYON feel native in every coding agent your customers use, from hooks to MCP gateways.

  • Ship and maintain integrations for Claude Code, Codex, Cursor and more
  • Define how new agents get supported within a week of launch
  • Mentor engineers on protocol design and backwards compatibility

Find the attacks before attackers do, then turn each finding into a detection that ships to every customer.

  • Red-team agents through issues, docs, web pages and tool output
  • Build and maintain Warden's injection corpus and evals
  • Publish research the wider community can use

Watch the registries agents install from, and catch typosquats and hallucinated package names as they appear.

  • Triage suspicious npm, PyPI and crates.io releases
  • Tune Sieve's provenance and reputation signals
  • Write advisories customers can act on the same day

Design approvals a developer can answer in one tap and dashboards a CISO can read in one glance.

  • Own the approval flow across terminal, Slack and web
  • Shape the visual language of Logbook timelines
  • Test with real security and platform teams every sprint

Run the demo against a customer's own repo, then stay with them until every agent is covered.

  • Lead technical evaluations with security and platform teams
  • Write the policy packs new customers start from
  • Bring field feedback straight into the roadmap

Teach developers to build fearlessly with agents, and show them what secure-by-default looks like in practice.

  • Write guides, talks and sample policies
  • Run the open-source policy pack community
  • Turn real incidents into lessons, with permission
How we hire

Four steps. No surprises.

The same process for every role, so you always know what comes next and how long it takes.

1

Intro call

30 minutes

A conversation about your work, the role and what the watch looks like day to day.

2

Work sample

About 3 hours

A realistic task from the team you'd join, done on your own time. No trick questions.

3

Team loop

Half a day

Walk us through your sample, then meet the people you'd work with every week.

4

Decision

Within a week

A clear yes or no with reasons, and an offer you can compare line by line.

Don't see your post? Write to us anyway.

Tell us what you'd build and why it would keep agents safer. Send it to careers@raksyon.tech.

Read our vision