Security, from both sides of the boundary.
Field notes on defensive engineering, adversarial testing, OSINT, machine-assisted research, RF and the systems underneath them. I build things, put pressure on them, and write down what actually happened.
Not the End of the World
A root shell is usually treated as the end of the story. This project asks what a Linux system can still protect after userspace root is compromised—and tests the answer instead of trusting the design.
Read the full test reportOne lab, several lines of inquiry.
The common thread is evidence: instrument the system, test the boundary, preserve the failed runs, and separate what was demonstrated from what merely looked promising.
Make the compromise less final.
Kernel hardening, recovery paths, tamper evidence and controls designed for the point after ordinary userspace trust is gone.
Give the attacker tools. Measure the difference.
Exploit-development ladders, constrained agents and security harnesses that test capability instead of arguing about it in the abstract.
Turn public records into auditable context.
On-prem intelligence systems, source-grounded correlation and tools that keep a visible path from a claim back to its evidence.
Follow the signal all the way down.
Radio experiments, passive LTE observation and the less glamorous work of finding out whether a peak is really a decode.
The result includes the part that broke.
These are working notes from an independent lab, not a consultancy pitch or a credential reel. The useful part is often the gap between the first idea and the thing the test harness would finally accept.
- 01Start with a concrete question and a boundary that can fail.
- 02Build enough instrumentation to tell policy from coincidence.
- 03Attack the test, vary the route, and record the ordinary failures too.
- 04Publish what the evidence supports—and name what it does not.
A long-running lab notebook, not a polished résumé.
I am not presenting myself as a professional authority. I have simply spent a long time around computers, security people, strange networks and the kinds of questions that do not leave you alone.
Substrate Labs is where that curiosity becomes something inspectable: code, test results, field notes, wrong turns and the occasional system that survives contact with reality.