Kyle Protho
Risk Management / Open Source Research / Solution Builder
I reduce uncertainty for decision-makers by framing the right questions, judging the evidence, and directing the analysis AI now accelerates.
How I Work with AI
I use AI to move faster. I put my own effort into the parts it can't own: designing the question, judging the evidence, and validating the result.
Certified in Claude Code in Action, and completed DeepLearning.AI's Machine Learning Specialization and an AI in Healthcare certification — building a working understanding of how these systems work and where they fail.
Built LLM integration into Analysis ToolBox, my open-source Python library, alongside the statistics and risk-analysis functions AI is now good at writing.
My rule working with AI: I decide which questions are worth asking after consulting with stakeholders, whether the evidence collected holds up, and how much confidence to put behind the answer before it reaches a decision-maker.
Publications
Moving Money for Better: The Role of Data Analytics in Western Union’s Anti-Human Trafficking Initiative
Explored the application of financial data analytics in detecting and preventing human trafficking networks.
Assessing organizational structure and identifying potential new members of criminal organizations using Twitter networks
Analyzed social media network structures to identify potential criminal organization recruitment and membership patterns.
Solutions & Platforms
Products and platforms I've designed and built end-to-end — from architecture to deployment.
ProblemAMP Fitness needed a full site rebuild with a live class schedule, a way for trainers to manage their own schedules without a developer, and a brand refresh.
What I DesignedRebuilt the site end-to-end in Next.js, Supabase, and Vercel — live class schedule, a custom admin portal for trainers and classes, and a full brand redesign.
ResultLive since 2026; outcome metrics not yet available.
ProblemReusable statistics and risk-analysis code rarely explains why a method is valid — which matters more, not less, as AI makes it easy to generate code without understanding it.
What I DesignedAn open-source Python library — 500+ commits, 100+ functions — covering data collection, statistics, and risk analysis: anomaly detection, Monte Carlo simulation, geospatial clustering, and LLM integration. Every function ships with a Teaching Note explaining the reasoning behind it, not just the syntax.
ResultOrganic user growth on the analysistoolbox.com platform since its August 2026 launch.
Star Rating Risk Management Tool
Healthcare Quality Analytics
ProblemCMS and NCQA Stars performance targets are usually tracked as single-point estimates, hiding how much uncertainty leadership is actually operating under.
What I DesignedA predictive risk management tool using predictive modeling and Monte Carlo simulation, giving leadership a probabilistic view of performance-target risk instead of a point estimate.
ResultUsed by Medicare and Medicaid business leaders to set performance targets and gauge the probability of hitting them.
Professional Timeline
Senior Quality Improvement Analyst
“Specializing in predictive optimization and intelligent automation to enhance healthcare quality measures.”
- ▹Decision Support Under Uncertainty: Designed a risk management tool for CMS and NCQA Stars, using predictive modeling and Monte Carlo simulation to give leadership a probabilistic view of performance-target risk instead of a single estimate. Used by Medicare and Medicaid business leaders to set targets and gauge the probability of hitting them.
- ▹Process Redesign: Rebuilt a manual reporting workflow end-to-end in Power Automate, cutting processing time 96% (from 20 hours to 45 minutes) and saving $12,000+ per year.
- ▹Executive Communication: Used SQL and Python to turn complex performance metrics into insights leadership can act on.
- ▹Data Strategy: Drove quality initiatives through data visualization and analysis in Power BI and Excel.
Vice President, Specialist – AML Compliance & Controls
“Assisted in regulatory compliance initiatives to strengthen AML controls and risk mitigation.”
- ▹Control Testing: Evaluated AML control efficacy and scope against regulatory standards, using SQL and Python to test whether controls actually caught what they were designed to catch.
- ▹Risk Mitigation: Identified control deficiencies and partnered with business units on audit-ready fixes.
Strategy Analyst
“Focused on population health strategies and competitive intelligence.”
- ▹Population Health Strategy: Analyzed member claims data in SQL, R, and Python to identify which population health strategies were worth pursuing.
- ▹Market & Competitive Intelligence: Ran OSINT research to test clinical service ideas against market reality before they reached leadership.
Sanctions Analyst
“Conducted financial investigations to detect fraud and sanctions evasion.”
- ▹Investigative Judgment: Assessed flagged entities and suspicious payments against sanctions and fraud typologies, judging which patterns warranted escalation.
Research Analyst
“Provided market intelligence for the cryptocurrency sector.”
- ▹Intelligence Program Design: Built an OSINT-based competitive intelligence program from scratch and launched the ICO Alert Crypto Minute, a daily decision-ready brief read by hundreds of subscribers.
Market Intelligence Analyst
“Monitored competitors and reduced uncertainty on market conditions.”
- ▹Intelligence Program Design: Built a second OSINT-based competitive intelligence program to monitor competitors and cut uncertainty around market conditions in target countries.
AML Risk Assessment Analyst
“Analyzed geopolitical patterns to forecast financial crimes.”
- ▹Risk Forecasting: Analyzed geopolitical and crime patterns to forecast money-laundering and terrorism-financing risk across the countries Western Union operated in.
- ▹Open-Source Investigations: Investigated parties in flagged transactions and wrote findings reports assessing which leads held up.
- ▹Visual Analytics: Built an interactive Excel infographic covering every country Western Union operated in, to streamline how risk assessments got made.
Background & Philosophy
About Me
I am motivated by the challenge of creating frameworks that enable organizations to move from reactive to proactive.
AI can produce analysis in seconds now; the harder work is deciding which questions are worth asking, which evidence to trust, and how confident to be in the answer. That's where I put my effort — framing the problem, testing whether the underlying data and controls actually hold up, and putting a number on the uncertainty instead of a guess. I've built that judgment across two regulated fields, financial crime and healthcare quality, backed by formal training in intelligence analysis. My job is to make sure the analysis, however it's produced, is one a decision-maker can act on.