I collect Windows and Sysmon telemetry, write Sigma rules, and run controlled attack simulations to check whether the expected alerts appear.
These projects document my work with telemetry collection, Sigma rules and attack simulation. The Active Directory write-up records attack stages and observed logs; detection claims require matching logic and validation.
Detection evidence
Explore the validation workflow, alert evidence and detection rules.
Validation run
Sentinel BAS ran ten Atomic Red Team tests on my lab host and matched each one to the alerts my SIEM raised, by ATT&CK technique, host and timing.
70%of tests detected
7 detected, 3 missed
ExecutionT1059.001PowerShellEncoded PowerShell commandDetectedin 0 s
PersistenceT1136.001Local accountCreate a new local userDetectedin 0 s
PersistenceT1547.001Registry Run keysRegistry Run key persistenceDetectedin 0 s
PersistenceT1543.003Windows serviceInstall a new serviceDetectedin 4 s
PersistenceT1053.005Scheduled taskCreate a scheduled taskMissedNo alert
Defense evasionT1070.001Clear Windows event logsClear the Security event logMissedNo alert
Credential accessT1003.001LSASS memoryDump LSASS memoryDetectedin 0 s
DiscoveryT1082System information discoveryDiscovery command burstDetectedin 21 s
ImpactT1490Inhibit system recoveryDelete volume shadow copiesDetectedin 0 s
Replay of a recorded lab run, not a live scan. A detection required a matching alert within 60 seconds. The 7/10 result applies only to these ten tests on this lab host; it is not overall ATT&CK coverage.
The Kerberoasting miss exposed a rule that could never fire: it relied on TicketEncryptionType, which the pipeline never extracted.
ARGUS samples selected Windows persistence, process, privacy, USB, network and LSASS-handle signals. Coverage depends on the collector, permissions and optional helper; it is not a replacement for endpoint protection.
Case-study summary
Problem
Develop a Windows assistant with explicit separation between AI-generated proposals and permission to execute actions.
Architecture
Svelte 5 / TypeScript HUD → FastAPI REST and WebSocket service → routing, authorization, action-bound grants and registered skills; separate voice workers.
My contribution
Developing the assistant, V2 HUD, agent kernel, Windows skills, monitoring and backend security controls.
Tests
The case study references public source code and documentation. Current Windows runtime test results have not been independently verified.
Outcome
Public Windows source release with voice, desktop workflows, bounded agents, a 3D state view and documented setup.