Scientist · Inventor · Builder
Geometric foundations
for augmented intelligence.
Michael R. Amiri/Augmenting human intelligence through geometric foundations.
Research
What if robustness, generalization, and interpretability are governed by the same geometric property?
My research investigates how the spectral properties of neural network weight matrices govern trustworthiness across deployment contexts. The central question: if we constrain the geometry of learned representations, can we simultaneously resolve robustness failures, generalization failures, and opacity as manifestations of the same underlying pathology?
Under peer review, 2026.
The Thesis
The future of healthcare is augmentation.
We have turned our healers into data entry clerks. The cognitive load of technical compliance has stolen the physician's attention from the patient. This is a failure of system design.
So we ask: what if the system carried the burden? What if intelligence could shield the clinician from administrative noise, amplify diagnostic power, and still leave every decision exactly where it belongs?
That is what I build. Intelligence that serves the physician. Systems that absorb the weight, surface the insight, and defer to human authority. The clinician commands. The system defends.
Sovereign Privacy
Patient data stays in the physician's domain. Privacy is enforced through physics, through mathematics, through architecture. Data is processed in isolation, the clinician's judgment is augmented in real time, and nothing is retained. The physician holds the keys. The physician controls the boundary. Always.
Augmented Cognition
Documentation drafts itself for your review. Codes surface as suggestions you accept, reject, or refine. Context-aware reminders appear when something is missed: gentle, adjustable, yours to override. Every insight is offered. Every decision is yours. The physician always has the final word.
Human Authority
Every output is a recommendation. Every action requires explicit physician approval. Nothing writes back to the EHR without your command. The system amplifies your expertise. It protects your time. It defers to your judgment. You trained for this. You earned this. You remain the decision-maker.
The Ventures
Where the work ships.
A research-and-development company and a private practice, each with its own front door.
Research & Development
PhysOpsAI
Verification and assurance for AI in the physical world.
A verification-and-assurance R&D company for AI that operates where failure is quiet. Version-bound evidence, bounded claims, and the instruments and mathematics that turn a system’s reliability into a property you can check.
Private Practice
AMIRI XI
Private intelligence, installed and operated.
A confidential AI practice for principals: local-first systems built, installed, and operated on hardware the client owns, capped at three principals a year. The same discipline runs a dedicated vertical for clinical practice owners.
The Journey
Education & Experience
M.S. Computer Science
Georgia Institute of Technology
Specialization: AI, Computational Perception & Robotics
2025 – Present
Bachelor's Degree, Computer Science & Data Science
NYU Courant
School of Mathematics, Computing, and Data Science
2023 – 2025
Podcast
AMIRI AXIOM
Conversations at the edge of AI, healthcare, and systems thinking. Grounded in first principles.
Community
United Teams
The Intelligence Exchange Protocol.
High friction. Zero noise. A filtered network for builders, researchers, and architects where intelligence density is the only metric.
Learn more at UnitedTeams.ai →Channels of Communication
Select the appropriate frequency. Whether for high-level consulting, professional networking, or intellectual exchange.
Ways to Work Together
Expected Response Time: 24-48 Hours