← Back to projects & research
Detection engineering2026

Detection-as-code pipeline

Built a Sigma pipeline that validates detection rules and compiles them to KQL for Sentinel and Defender on every commit.

My role

Author and maintainer.

Tools & technologies

Sigma · KQL · MITRE ATT&CK · GitHub Actions · Sentinel · Defender

Outcome

Four detections mapped to MITRE ATT&CK: password spray (T1110.003), brute force (T1110.001), malicious inbox rule creation (T1098), and impossible travel.

01 / Context

The problem

Detection logic written as one-off SIEM queries is unreviewable and quietly rots. It needed to live in version control with the same gates as application code.

02 / Implementation

What I built and owned

Detection validation workflow
Sigma rulesVersioned in Git
GitHub ActionsValidate & compile
KQL outputDetection queries
Sentinel
Defender

Rules are validated and compiled on every commit. Each detection includes false-positive analysis and validation steps.

Review. Validate. Compile.

Sigma rules live in Git. GitHub Actions validates their syntax and compiles them into KQL for Microsoft Sentinel and Defender on every commit. Each rule includes false-positive analysis and validation steps.

A focused detection catalogue

Four detections mapped to MITRE ATT&CK: password spray (T1110.003), brute force (T1110.001), malicious inbox rule creation (T1098), and impossible travel.

03 / Sources & links

Explore the work

Detection-as-code pipeline source (opens in a new tab)Detection catalog source (opens in a new tab)
← Return to résuméNext projectPrompt injection research