Offensive Researcher

Offensive Researcher

Tel Aviv | Full-time | On-site

Attackers have AI now.

Jailbroken frontier models, purpose-built uncensored ones, and agents that operate at machine speed.

The gap between a low-skill actor and a capable one is closing fast.

Our customers stay ahead of that. You're how.

PTBOX lets security leaders build and automate tailored attack vectors aimed at what matters: their crown jewels.

Not a CVE list sorted by CVSS. Not a pile of misconfigurations nobody will ever chain together.

The question we answer is whether a real attacker, using current tradecraft, can reach the systems and data the business runs on.

We tell customers our toolkit keeps pace with every threat that emerges. Your job is making that true.

MEET BREACHER

Breacher is our own LLM.

Local, purpose-built for offensive work, and the core of how we research here. No cloud, and no hedging when you ask it to do the actual job.

You'll work with Breacher daily: directing research, generating, and iterating payloads, exploring attack paths at a scale manual work can't reach.

Your findings make Breacher sharper. A sharper Breacher makes your next find faster.

THE JOB

Explore.

Track the threat landscape and work out what real adversaries are going after, including how they're weaponizing AI.

Build the tools that don't exist yet.

Turn all of it into product: new tools in the platform, new attack vectors, new scenarios that go after crown jewels instead of padding findings count.

You'll also run penetration tests for our customers. Real engagements, real environments, and reports people act on. It keeps the research grounded in what holds up in the field.

As a bonus, the same scenarios reveal whether anyone on the defensive side notices.

Your research doesn't end up in a report. It ships.

YOU

• Real offensive experience with enterprises
• You think in attack paths and business impact, not vulnerability counts
• Able to run a client engagement end to end, from scoping to a report an executive can act on
• You write your own tooling (Python, Go, Bash), not just run other people's
• Strong fundamentals: network security, OS internals, MITRE ATT&CK
• Comfortable with Burp Suite, Metasploit, and the rest of the standard kit
• Up for working with AI every day. No LLM background needed, just the judgment to know when a model is wrong
• The instinct to chase things nobody assigned you

BONUS

OSCP, OSEP, or CRTO.

Published CVEs, research, or open-source tooling.

Local LLM or agent experience.

Purple team background.

Reverse engineering.