Adrian Cisneros Contact

Research

Experiments and papers, with the real numbers and the parts that did not work.

Paper 46, explainedAn interactive explainer of my gesture radar paper, for people who have not read it.A line chart of detection score for six detectors as up to 20 percent of stored numbers are set to zeroA fly-brain detector for factory control attacksA pilot test of a brain-inspired attack detector for factory controllers on failing hardware.
A flow diagram of one research day in the agent town, where sealed held-out data feeds only the replicatorA town of AI agents looking for brain findingsA survey and a prototype town of AI agents that test brain-data claims on sealed data.The headline I scanned 2,487 AI-built apps above a code block with a policy set to trueI scanned 2,487 AI-built appsA security scan of 2,487 public AI-built apps found databases anyone could write to.
Adrian holding a printed copy of the journal Military Cyber AffairsAI that keeps working on damaged chipsI tested an AI that keeps working when the chip it runs on starts to break.A line chart of validation loss across head widths from 4 to 128 for five normalization variantsSet the temperature, not the normA small study showing attention normalization mostly works by setting the starting scale.
Adrian presenting his research poster to visitors2nd place for my research poster2nd place in the student research poster competition at the VICEROY Symposium, William & Mary.Teaching a gesture radar a new user without backpropagationOctober 2026A short paper for IEEE-EPIC 2026 on retraining a radar gesture model with forward passes only.
Ten problems in AI right nowOctober 2026A sourced survey of ten areas of AI risk, with fixes and gaps for each.Atria, a research agent that tries to prove itself wrongSeptember to October 2026I set an AI research agent on one idea. It wrote three papers and corrected itself.
Reproducing a Meta AI paper on free GPUsJune to August 2026I reproduced Meta's COCONUT paper on free GPUs. I got 95.33 percent against 97.0.NetState, a test of whether AI can track a changing networkMarch 2026A generated benchmark that checks if AI can keep track of a network under attack.
AriadneJanuary 2026I compared two kinds of AI on reasoning puzzles. The standard model failed longer ones.Ambient Contextual AINovember 2025A private assistant that knows what is on your screen without sending anything online.
HyperionNovember 2025An experimental AI memory you can edit directly, with no retraining.Itera-LiteOctober 2025A small state-space AI model I shrank 2.27 times with mixed precision.

Other kinds of work

Software and toolsApps, tools, and games I have built, from a hackathon win to software for the U.S. Air Force.6 projectsAI and automationSystems that do real work on their own, and the tools I use to build and test them.8 projectsWebsitesSites I have designed, written, built, and hosted.7 projects
Film and videoShort films I direct, and videos my own software makes.8 projectsWorkshops and talksTalks and hands-on sessions for business owners, students, and conference rooms.4 projectsCommunity and leadershipThe groups I lead and the places I volunteer.5 projects

Want something like this?

Answer a few quick questions and see an estimate in about a minute.

Get an estimate

or email [email protected]