Chu Detection engineering & security operations

Role

Security engineer, Team 3 (Packet Patrollers)

Scope

Application & cloud security hardening for a capstone platform launching July 30, 2026

Leading the security workstream on a cross-functional capstone: closing out formal security findings, standing up dependency and dynamic scanning in CI, deploying a Wazuh SIEM with custom detection rules, and coordinating a pre-launch penetration test and audit, alongside cloud security controls built on the team's AWS environment.

What

An adversarial AI detection engineering system, built end to end

Why

To show that ML-based detection can be threat-modeled, hardened, and operated like any other production security control, not treated as a black box

Threat-modeled with STRIDE mapped to MITRE ATT&CK, then built a full pipeline: a CICIDS2017 data pipeline over Kafka, an XGBoost classifier that measurably beat a rule-based baseline on F1, adversarial hardening across five attack types using IBM's Adversarial Robustness Toolbox, and a production FastAPI service with JWT/API-key auth, immutable audit logging, and an automated alert-to-containment pipeline, all shipped with GitHub Actions CI/CD.

What

An automated AppSec pipeline and SOAR platform, targeting OWASP Juice Shop

Why

Manual triage doesn't scale; this proves a severity gate and automated response can catch and act on findings before they ship

Built a GitHub Actions pipeline running Semgrep, Gitleaks, Trivy, and OWASP ZAP with a severity gate that blocks deployment on critical findings, then wired up Shuffle SOAR locally with three action-oriented playbooks covering deployment blocking, developer identification, and CVE enrichment via the NVD API. Published a live dashboard tracking results in real time.