Qira LLC · Phoenix, AZ & Littleton, CO

Intelligent systems for complex networks.

A research-driven technology company applying a single network-coordination framework across machine learning, traffic intelligence, and cognitive science — built from scratch, validated on real data, and protected under U.S. patent.

Explore the research → Partner with us

U.S. Provisional Patent #64/002,166 Google TPU Research Cloud NYU faculty engagement Founded 2024

Tier 1 — Core R&D

Research & flagship systems

Two flagships, one thesis. LOLM is the intelligence — a language model built from scratch that measures its own uncertainty and issues a receipt for every turn. QEV is the proof — the encrypted, signed, tamper-evident container those receipts travel in, and a complete product in its own right.

5 flagship systems

Machine Learning · Flagship ★ Flagship

LOLM — Latent Order Language Model

Qira's flagship system: a language model built entirely from scratch — a hybrid Transformer–SSM architecture that separates language into surface and latent representations, rather than the standard single-stream approach.

Trained and reproduced across standard datasets on both GPU and TPU, supported by the Google TPU Research Cloud — the system the rest of Qira's research is built to serve.

Explore LOLM → Try the live model

Security · Flagship · Open Source ★ Flagship

QEV — Qira Encryption Vault

Qira's trust layer: encrypted, signed, tamper-evident envelopes for any artifact — browser, CLI, desktop, and Android — with policy-aware workflows and verifiable trust semantics. Argon2id, XChaCha20-Poly1305, Ed25519, SHA-256.

It is what lets a LOLM receipt become evidence: any run can be sealed into a real QEV envelope and re-opened by the public verifier, because a claim about what an agent did only counts if it cannot be quietly rewritten afterwards. QEV also stands entirely on its own: MIT-licensed, live across four platforms, and useful for any file that has to survive being doubted.

Open the platform → Desktop & Android

Traffic Intelligence Live Production

Phoenix Traffic Intelligence

Real-time intelligence for Phoenix freeways — corridors monitored around the clock and modeled as a coupled-oscillator network that predicts congestion cascades before they form.

Validated against the standard METR-LA research benchmark, with cascade predictions and crew-dispatch guidance generated continuously from live AZ-511 and HERE routing data.

View live dashboard →

Cognitive Science Live Study · Preprint

EGC — Expression-Gated Consciousness

A mathematical framework for the gap between what people know and what they can express under pressure — one gating function, g(K)=4K(1−K), and three confirmed response types.

Compressors, Expanders, and Suppressors — an empirical study with real subjects. Preprint published on Zenodo; ArXiv endorsement pending.

Take the study → Read the preprint

Developer Tools Live

Codey

The coding AI that sees your codebase as a network. Not autocomplete — structural intelligence that understands how every file, function, and dependency connects to everything else.

Qira's network-coordination thesis applied directly to software: treat the repository as a graph and reason over its structure, not just its text.

Visit Codey →

Tier 2 — Applied & Experimental

Products & experiments

Smaller bets, shipped tools, and side projects — built with the same rigor, but separate from the core research program.

3 shipped projects

AI Infrastructure Live · MVP

AEX — AI Efficiency Exchange

Measurement, verification, and registry for AI energy efficiency and compute flexibility. Workload identity, boundary-aware energy attribution, baselines, AES scores, sealed evidence, and demand-flex simulation — not a carbon marketplace.

Open AEX → Standalone deploy

Digital Commerce Live · Launching 2026

My Digital

A marketplace for verified digital goods. Every purchase issues a cryptographically signed receipt and license that anyone can verify — secured by Qira's QEV trust engine.

My Digital → Verify a receipt

Developer Tools Live

Oort

The prompt stack for solo developers — bring your own key across every major model, ranked by real usage. One workspace, every provider.

Visit Oort →

Approach

One framework, applied everywhere

Qira's research centers on a single insight: the mathematical structures that govern phase transitions in physical networks also describe critical behavior in traffic systems, cognitive expression, and language-model training dynamics.

The same mathematics that detects a freeway congestion cascade also identifies the latent-order dynamics inside a language model and the gating behavior of conscious expression. This is not analogy — it is the same theory, applied at different scales.

By grounding every system in network coordination theory rather than domain-specific heuristics, our models don't merely fit data — they capture the dynamics that generate it.

◇ Network Physics

Mathematical modeling of how information, congestion, and instability propagate through connected systems — Kuramoto coupling, phase synchronization, and operating-band regimes.

⟲ Real-Time Inference

Continuous monitoring loops that detect anomalies and predict state changes before they cascade — production systems that run around the clock.

⇄ Cross-Domain Transfer

A single theoretical framework — validated empirically across transportation, cognition, and language — rather than three disconnected tools.

Team

Founders

A two-person research company — theory and systems, working end to end from first principles to live deployment.

Brandyn Leonard

Co-Founder & Research Lead

Littleton, Colorado

Lead architect of LOLM and originator of the gating functions and boundary conditions that underpin Qira's network-coordination theory. Drives the mathematical architecture connecting traffic dynamics, cognition, and adaptive training, and manages large-scale TPU training.

Email GitHub LinkedIn Profile

Bryan Leonard

Co-Founder & Lead Researcher

Phoenix, Arizona

Systems architect behind Qira's live platforms — data infrastructure, real-time monitoring, and empirical validation. Leads deployment of the Phoenix Traffic Intelligence system and all of Qira's production engineering and cloud operations.

Email GitHub LinkedIn

Recognition & Validation

Backed by real credentials

Independent work, patent-protected and externally validated.

Intellectual Property

U.S. Provisional Patent #64/002,166

Network coordination methods. Filed March 2026.

Compute

Google TPU Research Cloud

Granted access for large-scale ML training, 2025–present.

Academic

NYU Faculty Engagement

Invited discussion on network-coordination frameworks.

Publication

Zenodo Preprint

EGC framework published; ArXiv endorsement pending.

Contact

Work with us

We partner with transportation agencies, research institutions, and technology teams that need rigorous, real-time intelligence for complex systems.

bryanleonard@imagineqira.com

Phone

(480) 256-9684

Elsewhere

GitHub · LinkedIn · imagineqira.com

For DOT & government agencies

Traffic intelligence, pilots, and operational deployment

→

For researchers & universities

LOLM, EGC, and network-coordination collaboration

→

General & partnership inquiries

Everything else — say hello

→