Skip to main content
Track Record · Background

The foundation behind the mission.

Twenty-four years spent on one problem: how do you build a model that is honest about what it knows and doesn’t know? Training in mathematics and physics, an applied-mathematics MS, and a career in model validation and calibrated uncertainty all circle the same discipline. That work now runs on de-identified longevity-cohort data, under a strict local-only governance posture, aimed at Alzheimer’s and ALS.

← Track Record

Who is behind this

Full legal name: Michael Key. The links below are owner-supplied verification handles.
CV
Available on request

The case for data access

The work that bears directly on handling sensitive health data is the de-identified longevity-cohort analysis behind this mission — the NHANES and HRS analyses — carried out under a strict local-only governance posture: data stays on controlled infrastructure, nothing sensitive is shipped to third-party services, and every claim is traceable to its source. The specifics of how that data is handled are documented on the data stewardship page.

Behind that work is the foundation it rests on: training in mathematics and physics, an applied-mathematics MS, and a career spent on model validation, uncertainty quantification, and fidelity trade-offs. The discipline that matters here is the discipline to state confidence intervals that are right, not impressive — a model that says “I don’t know” when it doesn’t know, and calibration that means something beyond a leaderboard number.

Honest neurodegeneration prognosis needs exactly that. Everything built on the molecular track — the conformal prediction intervals, the group-conditional calibration, the honest accounting of where the model is blind — is the direct application of it. This is the same question carried forward, aimed at a more important problem: the mission →

A career built around one question

For twenty-four years I’ve built models that people use to make high-stakes decisions — not in a research lab optimizing benchmarks, but in the field, where the model has to be good enough for someone to stake resources, missions, and sometimes lives on the answer it produces. That constraint changes how you build, validate, and defend a model.

The formative chapter was at MITRE’s Cross-Cutting Urgent Innovation Cell (CUIC), where I served as Chief Analyst — solving problems that cut across the entire Department of Defense through architecture-level trade studies that integrated sensors, platforms, networks, and decision timelines into a single coherent picture. Every one of these was a simplified but faithful model of a complex system, built to answer a question that mattered.

That’s where I developed Analysis Driven Modeling: the analysis question determines which model to build, at what fidelity, with which metrics. Not “here’s a model, let’s find a question for it.” The question always comes first.

Before MITRE, I spent seven years running my own firm. That experience taught me things you can’t learn inside a large organization — finding the work, winning it, and standing behind every answer. When you are the firm, the model has to work and the briefing has to land. After that, you never hand over a number you can’t stand behind.

Today’s AI work is rediscovering what the modeling and simulation community learned decades ago: the hardest part isn’t the model itself. It’s knowing how good the model needs to be for the decision it’s supposed to support. A model that’s impressively complex but not validated against the right metrics is dangerous, not useful — it creates false confidence. The same discipline carries straight over: stating intervals that are honest about what the data can and can’t support, and being explicit about where the model is blind — which is what neurodegeneration prognosis needs.

Where I’m taking this — the vision →

The full track record

A career building models across defense, aerospace, intelligence, and education. Every role added something different; the through-line is the same. (Most recent first.)
  1. Major Aerospace Prime — Space Division
    Senior Analyst, Space Systems
    Mission analysis and simulation for space systems. Trajectory optimization, mission design, and decision-support tooling for space operations. Applying the same fidelity rigor to orbital mechanics and sensor physics that was developed for multi-domain defense analysis.
    The Model: orbital mechanics, sensor physics, decision-support systems
    Space systems analysis Decision-support tooling Current role (as of 2026)
  2. The MITRE Corporation — CUIC
    Chief Analyst, Cross-Cutting Urgent Innovation Cell
    Led multi-domain analysis teams solving problems that cut across the entire Department of Defense. Effects chain modeling, architecture trade studies, and kill-chain analysis using AFSIM and custom simulation tools. Managed 20+ analysts across concurrent studies. Developed the Analysis Driven Modeling methodology: the analysis question determines the model, not the other way around.
    The Model: multi-domain operations, effects chains, architecture-level trade spaces
    20+ analysts managed DoD-wide studies ADM methodology created
  3. Independent Defense Consulting Firm
    Owner & Principal Analyst
    Founded and operated an independent defense analytics firm for seven years, delivering threat analysis, simulation, and systems engineering through subcontract and consulting work with MIT Lincoln Laboratory, CSC, Torch Technologies, and L-3 Coleman Aerospace.
    • MDA (longest engagement): threat engineering, architecture-level BMDS analysis briefed to MDA leadership, trajectory and signature modeling validated against intelligence benchmarks
    • JIEDDO: lead systems engineer for a counter-IED program; lead author on the system requirements document for a cognitive signature-matching system
    • Additional sponsors: ARDEC, AMCOM, NRL
    The Model: threat trajectories, missile signatures, BMDS architecture, counter-IED systems
    MDA, ARDEC, AMCOM, NRL, JIEDDO Multiple performance awards 7 years as business owner
  4. Booz Allen Hamilton
    Program Manager & Data Analyst
    Program management and quantitative analysis for defense and intelligence community clients. Statistical modeling, data visualization, and decision-support tools. Where I first saw the scale at which large organizations use analytic work, and how often the model is fine but the communication fails.
    The Model: statistical models, decision-support analytics
    IC and DoD programs Statistical modeling
  5. Teledyne Brown Engineering
    Defense Analyst
    My entry into defense modeling and simulation. Missile defense analysis, trajectory modeling, and physics-based simulation at one of Huntsville’s defense engineering firms. The transition from teaching to defense analysis, where the models have to be right because the consequences are real.
    The Model: missile trajectories, physics-based defense simulations
    Missile defense analysis Entry to defense M&S (modeling & simulation)
  6. Math Teacher
    8th Grade · Community College · Professional Tutoring
    Started as a math teacher and community college instructor. Teaching taught the most important skill in any of this: making sure the person you’re explaining to actually understands what you’re telling them and can act on it. An 8th grader who doesn’t understand your explanation doesn’t politely nod. They stare at you, and you have to find another way in. That skill matters more than any technical credential in every role since.
    The Model: pedagogical models of student understanding
    Foundation in education Communication as a core skill

Qualifications & background

Degrees and technical background, education first.
MS Applied Mathematics, UNC Wilmington
Fluid Dynamics, Probability & Statistics
BA Physics · BS Mathematics, UNC Wilmington
Mathematical Physics & Operations Research
Chief Analyst, MITRE CUIC
Cross-cutting national security analysis across the Department of Defense
7 Years as Business Owner
Independent defense consulting firm — built from scratch, delivered for DoD sponsors
Former Math Teacher
Middle school, community college, professional tutoring — the foundation of clear communication
Held an active security clearance (as of 2026)

Tools & methods

The technical stack behind a career in modeling, from defense simulation suites to modern machine-learning frameworks.

Simulation

  • AFSIM
  • STK
  • STAMP
  • POST II
  • DICE
  • TGx

Programming

  • Python
  • MATLAB
  • SQL
  • PyTorch
  • TensorFlow
  • JavaScript

Methods

  • Monte Carlo
  • Reinforcement Learning
  • Bayesian Inference
  • Optimization
  • Uncertainty Quantification
  • Conformal Prediction

Visualization

  • Interactive Dashboards
  • 2D/3D Graphics
  • Data Visualization
  • Canvas API