WALKER E.
ROBINSON
Building at the intersection of AI, policy & capital.
Three worlds, one translator.
“You can't meaningfully shape something you don't understand.”
AI is reshaping power, security, markets, and truth itself. The future it builds could be extraordinary or catastrophic. The difference comes down to how we fund, build, and govern it.
I work where three worlds meet and rarely understand each other: policy, technology, and capital. I study the national security implications of emerging tech. I build with AI. I work to understand how capital decides which technologies actually reach the world.
Policy & National Security
Researching how AI reshapes security and power. Published analysis on AI competition, governance, and autonomous systems.
Building with AI
Founder of Venator Intelligence, building and shipping multi-agent AI products.
Capital & Markets
Studying how capital decides which version of the future gets built, and building investment intelligence for the people who allocate it.
The Builder Record
I ship. Systems I have designed, built, and put in front of real use.
DeadReckoning
An open-source intelligence instrument that tracks global GPS jamming through publicly broadcast flight data.
Every airliner continuously broadcasts its own navigation integrity. Treated as a sensor network, the world's airline fleet reveals where GPS interference is happening. DeadReckoning aggregates roughly 24 million position reports a day from open ADS-B data into a longitudinal archive: chronic interference zones, sudden flare-ups, and how both move with world events. Each region carries its own statistical baseline, named watch regions carry analyst context, and the methodology states plainly what the instrument can and cannot see. Attribution is never asserted, only sourced.
Machine Politics
A tracker of where every UN member state stands on autonomous weapons, built for the Seventh Review Conference of the Convention on Certain Conventional Weapons in Geneva, November 2026.
A globe unfolds into a flat Equal Earth map of all 193 states, with views for each year's General Assembly vote, each state's stated position, and where the two disagree. Strands traces every state from its 2023, 2024, and 2025 votes to its position, and the Assembly, a 193-seat hemicycle, rearranges as you step through the resolutions. Every reading quotes a dated primary source and follows a published rubric. Nothing publishes until it is approved: a validator enforces the evidence rules, 169 tests guard the build, and CI checks the live site against a fresh build. Each position pairs a color with a texture, a full table serves keyboard and screen-reader users, and the data is free to download under CC BY 4.0.
Version Control
A public record of AI model releases from twelve labs since ChatGPT launched in November 2022.
Each entry shows when a model shipped, how the public can use it, and which documents the lab published: system card, model card, usage policy, paper. Every benchmark score says whether the lab reported it or Epoch AI measured it, and every number links to its source. The overview has Ctrl K search and a lab-to-lab compare. A scroll story walks through the record chart by chart, and an explorer lists every entry with its sources and archived copies of lab pages. A nightly pipeline finds new releases and queues them for review. Nothing publishes until I approve it.
Venator Intelligence
Founder · Applied AI finance lab · 2025-2026
A 15-agent AI system for investment research.
Fifteen specialized agents cover company financials, market data, macro trends, earnings, and portfolio risk. A question is routed to the relevant specialists, their research runs in parallel, and the findings combine into a single sourced report. Every report then passes a correctness layer before delivery: deterministic arithmetic checks, internal contradiction detection, and grading against a structured quality rubric with two hundred criteria.
In Development
What I'm currently building.
Running Dark
Running Dark tracks the ships that don't want to be tracked.
It compares two witnesses: what vessels broadcast about themselves over AIS, and what the European Space Agency's Sentinel-1 radar satellites actually saw, using vessel detections published by Global Fishing Watch. Where the two disagree, it asks why. It cross-references those gaps against government sanctions lists from the US, UK, and EU, OpenSanctions vessel records, and ship identity histories to flag tankers that switch off their transponders, change flags, or meet at sea to transfer cargo. It also draws on DeadReckoning's GPS interference data, so it can tell a ship faking its position from one whose GPS is being jammed. The first version covers three chokepoints on the sanctioned-oil route: the Gulf of Finland, the Gulf of Oman, and the waters off Malaysia.
Backchannel
Backchannel is a hyperlocal security intelligence tool for the regions the big platforms cover badly.
Local sources report first; news outlets and governments confirm later. Backchannel listens to public Telegram channels and local news for the earliest signal, then checks it against slower, more authoritative sources: ACLED conflict data, ReliefWeb and NGO security reporting, government travel advisories and embassy alerts, and internet shutdown monitoring from IODA, Cloudflare Radar, and OONI. Every claim is graded on the NATO Admiralty scale, the standard analysts use to rate source reliability and information credibility, and corroboration only counts independent sources. The shutdown data does double duty: it tells a district that has gone quiet from one where the internet has been cut. It tracks public channels and reports, never individual people.
The written record.
China's AI Ambitions and U.S. Competition
“Artificial intelligence has the potential to transform economies, reshape societies, and redefine what it means to be a global power.”
Innovation, Not Hardware: What DeepSeek Teaches Us About the AI Race
“Our success depends not on outspending the competition but on out-innovating them.”
Beyond a Race to AGI: Reimagining the Manhattan Project
“Safety should not be sacrificed in favor of speedy innovation.”
To Defend Taiwan, America Needs To Upgrade Its Defense-Industrial Base
“Should China inhibit U.S. access to the semiconductor market, via a blockade or an outright invasion of Taiwan, it would directly threaten our national security.”
Energy Is the Achilles' Heel of Taiwan's National Security
“Taiwan relies on imported fossil fuels for 97.7 percent of its energy supply, making it highly vulnerable to global supply chain disruptions.”
The path so far.
The Foundation
Learning how the world actually works.
Lehigh University
2018 - 2022B.A. International Relations
Undergraduate studies in international relations. Built a strong foundation in policy analysis, global affairs, and international security studies.
American University, School of International Service
2022 - 2024M.A. International Affairs
Graduate studies focused on U.S. Foreign Policy and National Security. Completed a DoD-sponsored capstone project with the Defense Security Cooperation Agency analyzing UAS deployment in Ukraine and Nagorno-Karabakh.
Finding the Mission
Where policy meets the bleeding edge of technology.
American Foreign Policy Council
2024Researcher
Authored policy papers on China's AI ambitions and U.S.-China technological competition. Published in The National Interest and Newsweek.
The Prometheus Security Project
2024 - 2025Founder & President
Founded an independent research initiative analyzing emerging AI's impact on U.S. national security. Led a team of seven graduate researchers producing policy analysis on AI safety and strategic competition.
Building at the Frontier
From analyzing the future to constructing it.
JusTech
2025 - 2026Strategy & Policy Officer
Developed AI compliance certification frameworks for autonomous weapons systems aligned with international humanitarian law. Supported international partnership development across LATAM and EMEA.
Venator Intelligence
2025 - 2026Founder
Applied AI finance lab. Designed and built a 15-agent AI investment research platform: fifteen specialized agents whose findings combine into a single sourced report, checked by a correctness layer before delivery.
The best conversations start with a hard problem.
If you're working on AI, national security, or what capital builds next, reach out.