Security Engineering & Tooling

Tuning noisy insider-risk alerts and building AI-assisted investigation workflows

I improved behavioral alerting, built Python workflows for repeatable evidence collection, created an AI agent that develops SPL with the analyst, and shipped Splunk dashboards with automated updates for leadership, HR, and Legal.

Technologies
Python Splunk SPL & Dashboards Exabeam GitHub Claude OpenAI

Impact & Outcomes

140 → ~26 Monthly Alerts

Behavioral baseline tuning eliminated alert fatigue and reduced noisy false positives by over 80%.

↓ 82% Fewer False Positives

Replaced brittle static signatures with contextual peer baselines and dynamic thresholds.

↑ TP True Positives Elevated

High-confidence anomaly triggers elevated genuine malicious behaviors for rapid triage.

Automated Stakeholder Briefings

Automated Splunk dashboards and Python pipelines keeping Leadership, HR, and Legal in sync.

The Challenge

Alert Fatigue & Brittle Detection Logic

Generic vendor detection rules triggered endless non-actionable alarms whenever engineers pushed code outside normal hours or cloned internal documentation, burying real insider threats in noise.

Manual Evidence Gathering Toil

Investigators spent hours manually writing repetitive SPL queries, normalizing mismatched timestamps across logs, exporting CSVs, and reformatting telemetry for executive briefings.

Engineering & Automation Architecture

01. Behavioral Detection Engineering

Refined detection logic in Splunk and Exabeam with contextual peer baselines, dynamic thresholds, and multi-signal confirmation. The updated baseline logic reduced monthly alerts from roughly 140 to about 26, cutting false positives while improving the rate of meaningful true-positive findings.

02. Python Investigation Pipeline Suite

Developed modular Python CLI scripts that ingest case identifiers, query SIEM APIs in parallel, normalize timestamps, enrich IP/user metadata, and generate standardized markdown/PDF case files automatically.

What I Shipped

Collaborative SPL Investigation Agent

Built an environment-aware AI agent that works with the analyst to turn an investigative question into Splunk SPL, explain its choices, and refine the query using the organization’s real indexes, sourcetypes, and field conventions.

Dashboards & Automated Stakeholder Reporting

Shipped Splunk dashboards for leadership, HR, and Legal reporting, with automated updates that kept the relevant stakeholders informed without requiring analysts to rebuild the same briefing by hand.