Cybersecurity
Black Wall: Neural Network Self-Defending OS
Code: ANX-CS-002
Investigate whether the operating system itself can host neural network models that watch, model and defend the machine continuously — a self-defending OS in which learned defence models are a first-class part of the kernel and the security policy.
Why it matters
Security tooling today sits on top of the OS and is blind to what happens below it. If the system could learn what its own kernel, processes and hardware behaviour look like when healthy, it could recognise an intrusion in progress rather than only the damage after it. Black Wall is the attempt to make the operating system defend itself.
Research gap
Where a learned defence model should live inside an operating system — user space, kernel space, or split across both — and how to run inference on kernel telemetry at syscall rates without adding unacceptable overhead remains largely open. Equally open is how such a model stays trustworthy when the very kernel it watches can be compromised.
Research questions
- 01 Where in the OS should a learned defence model live to see enough telemetry without hurting latency?
- 02 Can neural inference on syscall, scheduler and memory telemetry run at line rate on commodity hardware?
- 03 How does a self-defending OS detect that its own defence model has been tampered with?
- 04 What does a learned defence model add over classical kernel integrity and intrusion detection mechanisms?
Suggested technologies
Required skills
Team and supervisor
3-5 researchers, supervised by Muhammad Qasim.
Suitable for: Systems, AI, Cybersecurity, CS, Embedded.
Expected deliverables
- Design and prototype of a learned defence layer inside an operating system
- Overhead and detection benchmark against classical kernel security mechanisms
- Research paper on the boundaries and trust model of a self-defending OS
Work on this problem
Apply to the lab and this problem becomes your supervised research project.