Threat Intelligence & OSINT

Austin Smith

From physical reconnaissance to digital intelligence — I build automated pipelines that track online harms, verify facts, and force systemic transparency.

View My Work GitHub
15+ OSINT Repositories
r = 0.62 Core Finding (p < 0.001)
8+ yrs Security Experience

About Me

I'm a 29-year-old from Arkansas who applies reconnaissance, threat assessment, and security skills — trained across federal, state, and private security roles — to analyze complex data and track adversarial influence online.

I swore into the Army National Guard at 17, trained at Fort Benning as a Cavalry Scout (19D), and later worked in state corrections and private security (G4S). These roles demanded strict objectivity: observing environments, documenting facts, and separating actionable threats from background noise.

In late 2024, I noticed severe polarization and coordinated narratives dominating social media. The mainstream narratives from both sides felt like coordinated influence campaigns. I approached this the way I was trained — by tracking signals, not decoding content. I began teaching myself Python in October 2025 to automate this tracking, and within months built multi-agent AI pipelines processing unstructured data at scale.

My objective is to apply my physical reconnaissance background and AI engineering skills to track online harms, verify facts, and force systemic transparency.

What Drives Me

🎯

Evidence Over Narrative

Every finding ships with the scripts to verify it. Failed predictions are documented publicly — negative findings are data.

🔍

Accountability & Transparency

Maintaining strict analytical objectivity — documenting correlations without claiming causation, and openly correcting past errors.

🛡️

Protecting People

From military service to OSINT research, the mission is the same — identify threats before they become harms and make information accessible to everyone.

Technical Capabilities

📡 OSINT & Threat Intelligence

  • Open-source intelligence collection & analysis
  • Temporal pattern analysis & signal detection
  • Adversarial influence tracking
  • Multi-platform narrative monitoring
  • Source triangulation & verification
  • Threat assessment & risk documentation

📊 Quantitative Methods

  • Pearson & Spearman correlation analysis
  • Permutation testing & bootstrap methods
  • Granger causality & lag analysis
  • K-means & hierarchical clustering
  • Chi-square & Mann-Whitney U tests
  • Causal inference (DoWhy framework)

🤖 AI & Automation

  • Multi-agent LLM pipelines (Claude, Llama, Grok)
  • Automated web scraping (Scrapy, Zyte)
  • AI safety research & bias detection
  • Meaning drift detection systems
  • LLM-powered data extraction & parsing
  • Prompt engineering & verification layers

💻 Development & Data

  • Python (Pandas, SciPy, Statsmodels)
  • Flask web applications
  • Streamlit dashboards & Plotly visualization
  • Scrapy spiders & data pipelines
  • Git & GitHub (15+ public repositories)
  • Dataset design, cleaning & validation

Featured Projects

Live OSINT Dashboard ● Live

The Regulated Friction Project

An automated intelligence pipeline tracking temporal correlations between geopolitical "friction" events and institutional capital flows. Processed 8 years of data to identify a statistically significant correlation (r = 0.6196, p = 0.0004) between friction events and compliance responses with a 7-day median lag.

r = 0.6196 Correlation
p < 0.001 Significance
66 pairs Verified
PythonScrapyZyteLlama 17BClaudePerplexity SonarStreamlitPlotly
Live Dashboard Repository
Civic Tech & AI Safety ● Live

Bill Translator

A Flask web application solving legislative transparency — simplifying dense legal jargon while preserving exact legal intent to meet Arkansas Act 602's 8th-grade readability requirement. Features custom meaning drift detection that flags potential intent shifts automatically.

PythonFlaskAnthropic APITextstatNLP
Live App Repository
AI & Automation

The Speaker — OSINT ChatBot

A Bring Your Own Key chat interface powered by the Anthropic Messages API (Claude) with a bundled Knowledge Base. Users supply their own API key in the browser — the server never stores keys. Loads a structured OSINT knowledge base from markdown files at startup, prioritizing verified research over training data.

PythonFlaskAnthropic APICSRF ProtectionRate LimitingRender
Repository
Real-Time Intelligence

Live Trackers

A three-stage real-time monitoring pipeline tracking five active leverage nodes — Maxwell, Iran, Gulf SWFs, Oracle/Ellison, and Epstein Files. Uses Perplexity Sonar Pro for live intelligence, Llama Scout 17B for structured entity extraction, and local convergence detection to flag multi-node activity windows.

PythonPerplexity Sonar ProLlama Scout 17BStreamlitGitHub Actions
Repository
Signal Intelligence

UVB-76 Structured Signal Analysis

15-year empirical analysis of Russia's UVB-76 shortwave station transmission timing patterns. Used K-means clustering and permutation testing (p < .0005) to identify four distinct operational modes, revealing a shift from training-cycle patterns to event-driven signaling aligned with real-world military operations.

PythonK-means ClusteringPermutation TestingTime Series
Repository
AI Safety & Online Harms

AI Manipulation OSINT Case Study

Quantitative documentation of selective emphasis in AI-assisted OSINT research. Identified a 207:1 keyword-mention disparity across AI platforms during a two-month investigation, with cross-platform verification and documented asymmetric guardrail application. Includes raw data and full methodology for independent verification.

PythonPandasCross-platform AnalysisAI Bias Detection
Repository
Geopolitical OSINT

Sovereign Capital Audit

Mapping the structural dependency of the US Defense Industrial Base on Gulf Sovereign Wealth (Mubadala/PIF) and East Asian Industry. Verified OSINT assessment tracking capital flows, supply chain vulnerabilities, and national security implications.

OSINTFinancial AnalysisSupply ChainGeopolitics
Repository
Data-Driven Policy Analysis

State Policy Analysis

Examines how U.S. states respond to the data center and energy infrastructure boom — from regulatory accommodation to moratoriums. Statistical analysis of 42 federal funding events across 27 states with transparent correction of prior methodological artifacts.

PythonSciPyDoWhyPandasStatistical Testing
Repository

Experience

Oct 2025 — Present

Independent OSINT Researcher & AI Engineer

Self-Directed

  • Built automated multi-agent intelligence pipelines processing unstructured data at scale using Scrapy, Zyte, Llama 17B, Perplexity Sonar Pro, and Claude
  • Developed live OSINT dashboard (Streamlit + Plotly) fed by automated LLM extraction pipelines scraping the Federal Register
  • Identified and documented statistically significant correlations (r = 0.6196, p = 0.0004) between geopolitical events and institutional positioning across 8 years of data
  • Published 15+ public repositories with reproducible findings, open datasets, and transparent methodology — including openly documented failed predictions
  • Conducted quantitative AI bias research documenting a 207:1 keyword-frequency disparity across AI platforms and asymmetric guardrail application
  • Built a legislative readability tool (Flask + Anthropic API) with custom meaning drift detection, deployed live on Render
  • Developed a BYOK OSINT chatbot (Flask + Claude) with a bundled knowledge base, security-hardened session management, and rate limiting
  • Built a real-time leverage node monitoring pipeline (Perplexity Sonar Pro + Llama Scout 17B) with automated convergence detection running via GitHub Actions
Security Career (2014 — 2024)

Security & Reconnaissance Professional

US Army National Guard · State Corrections · G4S Security

  • Cavalry Scout (19D), Army National Guard — Enlisted at 17, trained at Fort Benning in reconnaissance, surveillance, and threat assessment in high-stress environments
  • Correctional Officer — Monitored and documented behavioral patterns in a controlled high-security environment, developing observational discipline and situational awareness
  • G4S Private Security — Physical security operations requiring strict protocols, environmental monitoring, and incident documentation
  • Core competencies across all roles: observational discipline, objective threat assessment, fact-based documentation, and separating actionable intelligence from noise

Get in Touch

I'm actively looking to join a team where my threat intelligence background, quantitative research skills, and AI engineering experience can contribute to meaningful impact against online harms.

github.com/Leerrooy95 Live Dashboard