AI Security Explained #1 - Sleeper Agents in Your AI Model Supply Chain
One trigger phrase flips a trusted AI model from safe code to SQL injection. How poisoned LoRA adapters hide backdoors, and how researchers find them in the weights.
Senior Windows Kernel & Security Engineer at Microsoft
I'm Akash Trehan, a Senior Windows Kernel & Security Engineer at Microsoft. I write about kernel internals, VBS, and systems security.
Connect with me
Windows kernel internals, the Secure Kernel, VBS/HVCI, and how modern OS security actually works under the hood.
Reverse engineering and binary exploitation walkthroughs from CTFs — BackdoorCTF, OverTheWire, pwnable.kr and more.
Open-source contributions and side projects across systems, symbolic math, and developer tooling.
One trigger phrase flips a trusted AI model from safe code to SQL injection. How poisoned LoRA adapters hide backdoors, and how researchers find them in the weights.
How the Windows Secure Hotpatch Report uses VTL1 attestation to help anti-cheat and security tools tell legitimate hotpatches from kernel tampering.

What if sending email cost a digital stamp? Our blockchain hack won first place at HackInOut