nate@whitaker.net — zsh
80×24
nate@whitaker.net:~$ whoami

nate whitaker

cybersecurity · software · ai · polymath by practice

focus
cybersecurity · software · ai
role
code security engineer @ solventum
location
jacksonville, fl
experience
29 years
certs
cissp · lean six sigma black belt
status
available for conversations
$

cat about.md

// 01

I'm a multi-domain cybersecurity engineer with 29 years in the field — spanning mainframes, networking, compliance, vulnerability management, penetration testing, forensics, and application security. That breadth is the foundation; what I'm building on top of it is what matters now. My papers indicate that I am a CISSP and Lean Six Sigma Black Belt.

My current focus is software engineering and applied AI. I run the enterprise SAST program at a Fortune 500 healthcare company, where I've built a five-layer triage pipeline using local LLMs to cut analyst noise by orders of magnitude. I write production Python, design agentic systems, and treat models as components in real engineering — not magic. Where I'm heading: deeper into AI-native security tooling, autonomous triage and remediation, and the kind of local-first systems that don't ship sensitive data to someone else's cloud.

I'm a polymath by practice. I learn every new domain as if I'll have to teach it tomorrow — that constraint forces me to lock in fundamentals and build mental systems instead of collecting trivia. It's how I went from BBS-era hacker to sysadmin to network engineer to AppSec to AI systems builder, and it's why a problem in one domain often unlocks a solution in another. Outside the day job: writer, maker, black belt in Okinawan kempo, working English, conversational Spanish and French, enough Yiddish to make my grandmother smile.

$

ls -la ~/projects

// 02
threatbot-triage production

Five-layer SAST triage pipeline running local LLMs (Granite 4.0, Gemma 3, Mistral) via Ollama. Combines deterministic scoring, CxQL query overrides, multi-round agentic inference, and a human feedback loop. Processes 10,000+ enterprise scans/month and recovers hundreds of analyst hours.

python · ollama · checkmarx · xgboost · aws bedrock

companionbot in development

Local-first AI robot companion — no cloud, no telemetry. ESP32-S3 chassis, iPhone SwiftUI app, FastAPI personality backend running Granite models on dual RTX 3060s. Targeting DEF CON 34 prototype reveal.

esp32 · swiftui · fastapi · llama.cpp · ble

project twiggy scaffolding

Local-first ambient life-management platform built on the CompanionBot foundation. Privacy-respecting personal infrastructure for the post-cloud era.

edge ai · privacy · local-first

the land between novel

Historical fiction set in Civil War-era Tennessee. Multi-racial homesteading, the Battle of Shiloh, the Bloody Pond. Drawn from amalgamated family stories.

historical fiction · long-form

$

tail -n 5 ~/writing

// 03
— soon reducing sast noise at enterprise scale with local llms talk · def con 34 cfp
— draft the cxql override pattern: ground truth labels for ml triage technical essay
— wip why your analyst's job is a labeling pipeline essay
— ongoing the land between novel · in progress

// blog launching soon — check back

$

./connect.sh

// 04