Engineer · Researcher · Problem-Solver

I find problems
worth solving
and build teams
to solve them.

Kevin Williams — software engineer and AI researcher at NASA. I don't start with a technology. I start with a problem that matters.

See What Drives Me Get in Touch

01 — Why I Do This

Conviction First, Technology Second

Every project starts the same way. I see something broken, something hurting people, and I can't look away. The tools change. The reason never does.

An opioid crisis destroying communities

Building a Case for Safer Alternatives

I watched oxycontin tear through communities. When I got the chance to research at Michigan, I chose electrochemical THC detection — not because biosensing was trendy, but to build the scientific foundation for cannabis as a testable alternative to opioids. Give people a safer option and give society the tools to manage it.

Published Research0.13 μM LOD

AI repeating social media's mistakes

Making AI a Net Positive for Humanity

Social media promised connection and optimized for addiction instead. I see AI at the same crossroads. That's why I moved into AI safety and certification at NASA — not to slow AI down, but to build frameworks that make AI on a spacecraft carrying humans actually earn the trust we place in it.

NASA Silver Bear AwardNPR 7150.2D

Legacy systems slowing human spaceflight

Replacing Bureaucratic Friction with Real Tools

At NASA, teams wrestled with SharePoint workflows never designed for spaceflight complexity. Data was scattered, processes were manual. So I built Cortex — not because I love enterprise software, but because people doing the hardest work in aerospace deserved better tools.

Cortex Platform~20 hrs/wk Saved

02 — The Work

Problems I've Tackled

Spaceflight ops bottleneck

Cortex Platform

Django-based enterprise platform with EAV architecture replacing scattered SharePoint workflows for NASA human spaceflight ops. Non-technical users define custom data structures; automated pipelines turn unstructured inputs into validated data.

DjangoPythonEAVFlutter

No safety standards for AI in space

AI Certification Framework

First-generation certification framework for AI on human-rated spacecraft. Addresses bias detection, continuous validation, hallucination prevention, and human-AI teaming. Guiding AI deployment across Artemis mission elements.

AI SafetyNPR 7150.2DV&VGovernance

Astronaut monitoring vs. privacy

Cognitive Twin System

Privacy-preserving cognitive readiness prediction for astronaut monitoring. Implemented differential privacy and defended against membership inference attacks. Monitoring health shouldn't mean sacrificing privacy.

MLDifferential PrivacyBiosignals

Opioid crisis needed alternatives

Electrochemical THC Detection

Published research on Δ9-THC detection achieving 0.13 μM LOD — third-lowest among comparable devices. Built to support responsible cannabis legalization as an opioid alternative.

ResearchElectrochemistryPublication

90% of inspection time was manual

Deep Learning Quality System

At Ford, deployed deep learning for automated quality decisions in eMotor manufacturing. Reduced manual inspection by ~90%. AI frees people to do higher-value work.

Deep LearningManufacturingFord

03 — The Path