$ ./qore --init --secure · SOVEREIGN · AIR-GAP NATIVE · 2,100 TESTS PASSING

Your AI took your data and left.

Every other AI sends your data somewhere else to think. Qore doesn't. Autonomous agent teams, multi-model reasoning, and encrypted orchestration — running entirely on your hardware.

No cloud. No data egress. No per-token meter. Every byte stays in your custody. Backed by 2,100 automated tests. One install, and it works offline — indefinitely.

See the Platform Request a Briefing
root@qore:~ — qore boot
Deployable AI Models◆ Autonomous Agent Teams◆ Advanced Cryptography◆ Local Execution◆ Quaternion Neural Architecture◆ Edge AI Deployment◆ Post-Quantum Security◆ Encrypted Orchestration◆ Air-Gap Native◆ Self-Hosted Inference◆ Trusted Execution◆ Genesis Mission Ready◆ Deployable AI Models◆ Autonomous Agent Teams◆ Advanced Cryptography◆ Local Execution◆ Quaternion Neural Architecture◆ Edge AI Deployment◆ Post-Quantum Security◆ Encrypted Orchestration◆ Air-Gap Native◆ Self-Hosted Inference◆ Trusted Execution◆
Agentic Workforce
Autonomous agent teams with memory, tools, and delegation.
Runs on Your Hardware
Local, edge, or fully air-gapped. Zero cloud dependency.
Post-Quantum Secured
ML-KEM-768 + ML-DSA-65. Encrypted everything, by default.
Verified, Not Claimed
2,100 automated tests prove every security guarantee.
National Imperative · The Genesis Mission

The mission named the problem.
Qore is the answer.

The Genesis Mission established a national imperative: harness AI to accelerate American science and security — on infrastructure the nation owns and controls. Its hardest unsolved problem is not larger models. It is running frontier inference on the most sensitive data in the world without ever surrendering custody of that data. That problem has a name. It also now has an answer.

"How do you put frontier intelligence on classified, proprietary, and mission-critical data — without exposing that data at compute time, and without handing it to a hyperscale cloud?"

— The infrastructure question at the heart of the mission

Conventional AI decrypts before it computes. It depends on infrastructure no single nation controls. That is a structural mismatch with the mission's requirements — not an engineering backlog. Qompute AI was engineered from the ground up as the answer. Local execution, mathematically efficient quaternion inference, post-quantum transport, and client-owned data — a sovereign AI substrate built for exactly the environments the mission demands, already deployable today.

Sovereign Compute
No Hyperscale Dependency
The mission cannot run on infrastructure controlled by external parties. Qore runs locally, at the edge, or fully air-gapped — no cloud intermediary in the execution path, data never leaves client custody.
Efficient Intelligence
Frontier Capability, Commodity Hardware
Scaling can't depend on unlimited power and silicon. Published quaternion architecture delivers 98% accuracy at ~75% fewer parameters — advanced inference on hardware the mission can actually field.
Trusted Execution
Post-Quantum, Tamper-Resistant
Sensitive data must survive a quantum-capable adversary. Transport is post-quantum by design — ML-KEM-768 and ML-DSA-65; the host OS is treated as untrusted — built for environments where the network is hostile.
The Platform · What Qore Does

Not a chatbot.
An operating system for agents.

Qore is a complete, self-hosted platform — not a wrapper around someone else's API, not a library you wire together. Autonomous agent teams, multi-model reasoning, encrypted task orchestration, and distributed mesh collaboration, all running inside your boundary. Zero cloud dependencies. Every capability below ships today and is guarded by automated tests.

01
Autonomous Multi-Agent Teams
Agentic Workforce
Persistent AI agents organized into teams with shared objectives. Each has encrypted memory, its own tools, and a goal it works toward across runs — planning, executing, delegating to peers, and scoring its own output. Your agents get measurably better at their jobs over time.
02
Plan · Execute · Synthesize
Task Orchestration
Multi-wave execution plans with dependency graphs. RAG-grounded expansion for 10,000+ word deliverables. Self-healing validation that repairs failed output before it ships, and checkpoint/resume that survives interruption. Fully autonomous, start to finish.
03
Capability On Demand
Tool Forge
Agents build their own tools through a seven-stage pipeline — specified with Gherkin behavior profiles, then implemented, tested, validated, repaired, and Ed25519-signed before deploy. Nothing ships below a quality score of 7. New skills, quality-gated.
04
The Smart Model For Smart Work
Multi-Model Reasoning
Local, Metal-accelerated inference with a 131K context window. Phase-aware routing sends planning to the reasoning model and execution to the fast model — automatically, with load balancing and fallback. No API keys. No per-token billing.
05
Distributed, Not Centralized
Peer Mesh
Connect Qore nodes into a collaborative swarm with post-quantum-secured, trust-gated delegation and encrypted knowledge transfer. Send workflows to remote teams with cryptographic provenance — distributed AI with no cloud in the middle.
06
Client-Side Keys, Server Sees Nothing
Encrypted Everything
AES-256-GCM with HKDF-derived keys on canvases, tasks, agent memory, workflow transfers, and forensics — each with per-document salt and IV. A stolen database file is unreadable ciphertext. On license expiry, cached keys are erased within 60 seconds.
07
Your IP Stays Yours
Post-Quantum Licensing
ML-DSA-65 signed, ML-KEM-768 encapsulated, hardware-bound license tokens (NIST FIPS 203/204). A stolen token is useless on another machine; a single flipped bit is detected. Even with full source access, proprietary agents and assets stay locked.
08
Prove What The AI Did
Governance & Audit
RBAC on every agent, tool, and model — fail-closed, deny-by-default. An immutable, hash-chained, RSA-2048-encrypted forensics log. SIEM correlation with real-time alerting. Every action is attributable; every decision is recorded and exportable.
Built For
Security-Conscious Orgs
Classified, ITAR, defense, finance, healthcare, and legal — compliance without compromise. Air-gapped, hardware-bound, fully audited.
Enterprise Engineering
Internal AI with governance built in — RBAC-enforced agent teams, complexity-aware routing, and encryption on by default.
Consultancies & MSPs
Redistribute agent IP, manage client seat pools, and retain the commercial relationship through a central License Authority.
Independent Developers
A local AI workbench that doesn't spy on you, doesn't meter you, and gives you real agentic capability. Starts free.
Code Canvas · 28 languages Auto-RAG · drag-and-drop Multimodal · image · audio · PDF Admin Dashboard Multi-User + SSO (OIDC) MCP Server Registry Checkpoint / Resume SIEM + OSINT Alerting
Core Technologies

Built on
four principles.

AI models, trusted systems, and technologies designed for modern computing.

01
Artificial Intelligence
Deployable Models
Purpose-built models designed to support security, healthcare, investigation, and specialized analytical workflows.
02
Zero Trust AI Fabrics
Trusted Execution
Architectures designed to support secure access, controlled interaction, and trusted AI environments.
03
Advanced Cryptography
Secure Communications
Technologies designed to protect information, strengthen trust, and support modern computing environments.
04
Mission Systems
Focused Capabilities
Built for applications across defense, digital forensics, healthcare, and critical infrastructure.
Why It Matters · Production Reality

Demos are easy.
Production is not.

The reason most AI agents never leave the lab: they hallucinate, they drift, they leak, and they can't prove what they did. In high-consequence environments that's disqualifying. Qore was engineered to close each failure mode — not with a prompt, but with architecture.

Hallucination
Grounded & Self-Correcting
  • Every run self-scored 1–10 vs. its goal
  • Auto-revision when quality is below threshold
  • RAG-grounded output with citation prompts
  • Self-healing validation, up to 2 repair passes
Runaway Agents
Scoped & Fail-Closed
  • Sandboxed with only declared tools
  • Delegation depth capped, cycles detected
  • Overlap prevention stops duplicate runs
  • Deny-by-default RBAC on every action
No Audit Trail
Cryptographically Provable
  • Immutable, hash-chained forensics log
  • RSA-2048 + AES-256-GCM, tamper-evident
  • Every message, tool call & decision recorded
  • SIEM-exportable for compliance evidence
Data Leakage
Nothing Phones Home
  • All inference local — zero cloud calls
  • AES-256-GCM at rest, client-derived keys
  • Server stores ciphertext it can't decrypt
  • Keys erased within 60s on license expiry
Unreliable in Prod
Engineered to Hold
  • System Repair Agent auto-fixes failures
  • 2,100 tests guard against regression
  • Deterministic local inference, no silent swaps
  • Zero TypeScript errors, zero flaky tests
Vendor Lock-In
Sovereign by Design
  • Self-hosted, open model format (GGUF)
  • Portable, PQC-encrypted data bundles
  • No subscription to someone else's GPUs
  • Never trained on your conversations
How It Works

From bare metal to
sovereign AI in ten minutes.

One command, or one Docker image. No account, no API key, no external service in the loop. Here is the full lifecycle — install to audit.

01
Deploy
Install
One interactive command provisions everything — runtime, local models, PKI keys, and the encrypted database. First login creates your admin account; the full platform seeds automatically. Offline-capable from that moment on.
02
Configure
Govern
Define agents and teams, bind tools and models, and set RBAC policies and complexity-routing rules — all from a single admin dashboard. Personas are cryptographically bound to the servers allowed to serve them.
03
Operate
Orchestrate
Launch a task, or let scheduled and event triggers fire on their own. Qore plans, routes to the right model, executes tools, reflects on quality, and synthesizes a final deliverable — streaming every step in real time.
04
Verify
Audit
Every action lands in the encrypted forensics log and SIEM. Review quality scores, export tamper-evident evidence, and answer "what did the AI do, and why" with cryptographic certainty — for any auditor, at any time.
Use Cases · The Workforce in Action

22 autonomous agents.
Put to work.

Qore ships with a standing workforce of 22 specialized agents across five teams — triggered by cron, interval, webhook, or event, and able to delegate to one another. They run on your schedule, inside your boundary, and escalate to a human only when it matters. Here is what they do.

Security Operations
A SOC That Never Sleeps
Continuous posture scans, event-stream anomaly detection, threat-feed ingestion, and tool-integrity checks — running around the clock and escalating only real signal.
Security AuditorReal-Time AlertingOSINT IntelligenceTool Safety Auditor
Forensics Automation
Answers in Minutes, Not Days
Docker-isolated binary and malware analysis, IOC enrichment and attribution, and a cryptographic chain of custody on every artifact — full-spectrum investigation, self-hosted.
Forensic AnalystOSINT CoordinatorEvidence Custodian
Self-Healing Ops
Systems That Fix Themselves
15-minute health checks, automatic root-cause diagnosis and repair the moment something fails, then verification before the incident closes. Uptime without a pager.
System Health MonitorSystem Repair AgentModel Server GuardianRuntime QA
Compliance & Audit
Evidence on Demand
Every action written to an immutable, hash-chained log; configuration drift caught against baseline; instrumentation gaps closed automatically. Aligned to GDPR, EU AI Act, and SOC 2.
Config Drift DetectorTelemetry CoverageDocumentation Sync
Knowledge Operations
Institutional Memory, Curated
Knowledge-graph maintenance, RAG collection quality management, and auto-generated documentation — all encrypted at rest and never leaving your network.
Assimilation ProcessorKnowledge CuratorTutorial Generator
Your Own Agents
High-Stakes, Custom-Built
Compose your own agents and teams with declared tools, bounded memory scopes, a complexity tier, and any schedule — governed by RBAC and capped delegation, so autonomy never means uncontrolled.
CronIntervalWebhookEventManual
Industry Solutions · Where It Operates

Six domains.
One foundation.

Information Security. Healthcare. Critical Infrastructure. Defense. Investigations. One engine — Qore — pointed at the realities of each environment.

Defense
Built for Mission Execution
  • Decision support
  • ISR workflows
  • Secure systems
Medical
Precision Where It Matters
  • Surgical AI
  • Tumor analysis
  • Local execution
Gaming
Built for Platform Integrity
  • Behavioral analysis
  • Risk detection
  • System visibility
Forensics
Designed for Investigation
  • Incident response
  • Evidence correlation
  • Threat analysis
Energy
Built for Infrastructure
  • Resource optimization
  • Operational visibility
  • System performance
Labs
Where New Systems Begin
  • Model development
  • Applied research
  • Technology validation
Our Mission

Secure AI for
the real world.

The Genesis Mission named the hardest unsolved problem in AI for high-consequence environments: how do you run frontier inference on the most sensitive data in the world without surrendering custody of it? Conventional AI can't answer that. Qompute AI was built to.

Every platform is engineered from first principles for environments where the network is hostile and data must never leave client custody — local execution, post-quantum transport, and mathematically efficient inference on hardware that can actually be fielded.

Purpose Built
Technology designed to solve real problems.
Trust by Design
Security is built into every layer.
Engineered to Perform
Efficiency, reliability, and precision without unnecessary complexity.
Ready to Adapt
Built to evolve alongside changing requirements and emerging technologies.
Inside Qore · QAI-PROD-001

From model
to mission.

The same governed, auditable Qore engine — agentic workforce, orchestration, and post-quantum transport — pointed at specific high-consequence domains. Local models, threat response, edge analysis, medical imaging, and forensics, all built on one sovereign platform.

Local Models · Open-Weight
Qore Models
Deployable AI models built on advanced open-weight frameworks, engineered for efficient local execution across security, analysis, edge AI, and specialized operational environments. Structured across three tiers — Foundational, Informed, and Specialized Intelligence — and run entirely inside your boundary.
Open-Weight ModelsQuantizationAny HardwareAir-Gapped DeployEdge Inference
Automated Threat Mitigation
Qore Threat Mitigation
An AI-driven threat mitigation platform that detects, analyzes, and responds to complex security events in real time. Combines offensive and defensive intelligence into a single system that understands how attacks happen — and how to stop them.
Threat DetectionAI-Driven ResponseOffense + DefenseLocal Deployment
Post-Quantum Transport
Qore Secure Transport
Qore's cryptographic and transport layer for secure data exchange and trusted communication — post-quantum key exchange and signatures (ML-KEM-768, ML-DSA-65) built in, the network treated as hostile by default. The secure foundation beneath every Qore workflow.
ML-KEM-768ML-DSA-65Zero-TrustAir-Gap Native
On-Device Media Analysis
Qore Edge Analysis
Workload-specific media processing that reduces footprint while preserving usable clarity and enabling real-time analysis. Instead of treating video as raw data, Qore processes and analyzes content on-device — transmitting metadata results instead of full media when appropriate. One of Qore's most demonstrable capabilities, with working pipelines and live demos.
High-Ratio CompressionEmbedded AnalysisMetadata-FirstLive Demos
Medical Imaging
Qore Medical Imaging
Specialized AI for diagnostic intelligence — imaging analysis, cancer detection workflows, 3D tumor visualization, and surgical planning support. Designed to operate efficiently and deploy directly within clinical or research environments, bringing advanced analysis closer to the point of care.
Cancer Detection3D VisualizationEdge DeploymentSurgical Planning
Forensics & Incident Response
Qore Forensics & IR
AI-driven incident response and investigation: automated timeline analysis, a context-aware Case Agent, an investigation workbench, and evidence correlation across systems. Paired with governed offense-and-defense agents that simulate attacks and generate defensive strategy — full-spectrum coverage for operators who need answers in minutes, not days.
Timeline AnalysisCase Agent AIEvidence CorrelationRed + Blue Team
Why Qore · The Difference

Everyone else asks you
to give up something.

Hosted AI wants your data. DIY agent frameworks hand you a box of parts and a weekend of wiring. Qore is the integrated, sovereign alternative — the full platform, on your hardware, with nothing to surrender.

Replaces your hosted AI
vs. ChatGPT · Copilot · Bedrock
  • Your data stays local. Never sent, never used for training.
  • Persistent agent teams, not a stateless prompt box.
  • A full platform, not code completion — orchestration, multi-model, admin.
  • No per-token billing. One-time deploy, deterministic local inference.
Replaces your DIY agent stack
vs. LangChain · Open WebUI · Ollama
  • An integrated application, not a library you assemble yourself.
  • Workforce, Tool Forge, mesh & routing built in — production-grade, not a hobby project.
  • Governance included: RBAC, forensics, PQC licensing, SIEM.
  • Zero-config install with UI, admin panel, and desktop packaging.
Every Claim, Verified
2,100
Automated Tests
Full system verified in ~10s. Every commit must pass before merge.
ML-KEM · ML-DSA
Post-Quantum
Post-quantum algorithms (ML-KEM-768 / ML-DSA-65) across licensing and mesh.
AES-256-GCM
Encrypted at Rest
Canvases, tasks, agent memory, forensics — HKDF-derived, per-document.
0
Cloud Calls Required
Runs fully air-gapped after activation. Your hardware, your inference.
GDPR Art. 25 · 30 · 32 · 44 EU AI Act Art. 10 · 12 SOC 2 CC6 · CC7 HIPAA §164.312 NIST SP 800-53 CNSA 2.0

Security properties describe design objectives, verified by automated test coverage. Regulatory references indicate alignment, not certification. Benchmark figures reference the MNIST dataset and do not imply production or domain accuracy. Full methodology available to qualified parties under NDA.

Who Builds This · QAI-TEAM-CORE

Built by people who
have been inside the breach.

Qompute AI's core is engineered by operators from the front lines of incident response and adversarial research — the people enterprises and governments called when nation-state actors were already inside. This is not a team that read about threat models. They wrote the response playbooks.

DH
Derek Hinch
Chief Technology Officer

Derek Hinch is an artificial intelligence architect, cybersecurity engineer, and U.S. Air Force Electronic Warfare R&D, Operational Test and Evaluation Veteran with nearly three decades of experience designing, securing, defending, and penetrating large-scale distributed defensive systems and foreign materiel. His technical background spans the full spectrum of high-stakes computing, from engineering hyperscale cloud infrastructure to leading Fortune 10 enterprise security at Amazon where he was responsible for the core security of Amazon for the s-team.

Today, Derek bridges the gap between theoretical mathematics and tactical edge computing. He is the published author of the quaternion neural architecture that powers the Argus engine—a foundational breakthrough that drastically optimizes complex machine learning models for constrained environments. By fundamentally rethinking the underlying mathematics of neural networks, Derek’s work enables Genesis and programs like it to deliver uncompromising, mission-grade AI inference natively on ruggedized, field-deployable hardware, bringing heavy computational power directly to the tactical edge.

  • U.S. Military Electronic Warfare PRIOR
  • FAANG Cloud Senior Security Engineer PRIOR
  • Fortune 500 Enterprise Director of Security PRIOR
  • Global Security Firm Principal Consultant PRIOR
JV
Joey Victorino
VP, Infrastructure & Security
A practitioner-skeptic shaped by a decade on the front lines of real intrusions — ransomware, advanced persistent threats, and live nation-state activity — across elite incident-response practices at a global cloud giant, a top-tier endpoint-security firm, and a Fortune 100 technology consultancy. M.S. Cybersecurity. Leads the infrastructure, security architecture, and post-quantum research that makes Qompute AI's Genesis Mission substrate defensible by design.
  • Global Cloud Giant Elite Incident Response Team PRIOR
  • Top-Tier EDR Firm Sr. Cloud Security Consultant PRIOR
  • Fortune 100 Consultancy Principal IR Consultant PRIOR
  • Global Security Firm Sr. Security Consultant PRIOR
Qore Labs · Research Foundation · QAI-RES-001

The math is published.
The results are real.

Qompute AI's efficiency advantage is not a roadmap promise. It traces to peer-documented research on quaternion algebra applied to neural networks — benchmarked, reproducible, and the foundation of the Argus engine.

Published · August 2024 · Hinch-2024-QNN
Revolutionizing AI with Quaternion Algebra:
A Leap in Neural Network Efficiency
Derek Hinch — Founder & Chief Technology Officer, Qompute AI
98.04%MNIST Accuracy
0.083Test Loss
49KParameters
<2sTrain Time
Core Contribution
A custom QuaternionDense Keras layer replacing standard matrix multiplication with the Hamilton product in four-dimensional hypercomplex space — q = w + xi + yj + zk, where i² = j² = k² = ijk = −1. The non-commutative structure encodes rotational and orientational information more efficiently than conventional linear algebra.
Parameter Efficiency
98.04% test accuracy on MNIST with only 49,170 trainable parameters — versus ~200,000+ for a conventional dense network of equivalent depth. Training completes in under 2 seconds on a consumer laptop. The implementation is hardware-portable and available for review under NDA.
Why It Matters for Deployment
Quaternion operations map naturally to SIMD instruction sets — performance without floating-point overhead. The same efficiency that shrinks parameter count is what makes Qore models viable on consumer hardware, edge devices, and air-gapped systems without hyper-scale infrastructure.
From Research to Product
The published QuaternionDense layer is the foundational building block of the Argus inference engine inside Qore. Qore Labs continues extending the quaternion substrate to convolutional, attention, and embedding architectures — available for technical due diligence to qualified parties under NDA.
Qore Labs · Frontier Research · QAI-RES-002

The Q-Factor.
One geometry for security and size.

A research subsystem behind Qore's efficiency work. Its premise: the mathematics that makes data hard to break is the same mathematics that makes it cheap to store — so security and compression stop competing for the same budget. The early results are real and reproducible. They are also early. We show you both.

Qore Labs · docs/qfactor · verified 2026-06-28
Lattice-Quaternion Vector Store:
Unified Compression, Retrieval & Post-Quantum Integrity
Reproducible on an 8-document medical-knowledge validation corpus. At-scale benchmarks in progress.
~17×Smaller Store
32Dvs 1536D
100%Lossless Rebuild
8 docsValidation Scope
The Idea
Instead of storing each document as an isolated 1,536-number vector and recomputing relationships on every query, the Q-Factor stores the relationships themselves in shared lattice-and-quaternion geometry. The relationship lives once, in the structure — so the store is both non-redundant (small) and hard to pull apart (secure).
What Is Proven
The core transforms are provably reversible — identities like flux_unfold(flux_fold(x)) == x hold for all inputs and are test-verified, so reconstruction is exact, not approximate. Recovering a stored point without the key reduces to a lattice shortest-vector problem — the same post-quantum hardness family as ML-KEM / ML-DSA.
What Is Measured
On a small medical-knowledge corpus (8 documents, 5 queries): a 94% smaller store (≈17×) at a 32-dimensional embedding versus 1,536, with 100% top-1 retrieval and integrated tamper detection. Honest scope — this is a validation-scale result, not a production or domain benchmark. It is exactly the number we are now pressure-testing at scale.
What Is Projected — And Next
Because the storage ratio is a property of the algebra, it extrapolates linearly — a projection, labeled as one, pending large-corpus validation. The near-term plan is a benchmark at 10K–1M documents against a production vector database. Full methodology, code, and the deeper theory are available to qualified parties under NDA.

Q-Factor figures reflect a validation-scale corpus (8 documents); at-scale numbers are linear projections pending benchmarking. Reversibility and lattice-hardness properties are algebraic and test-verified. This describes active research, not a shipped product tier.

Infrastructure Specification · QAI-SPEC-001

The technical truth.
No marketing.

Everything engineers and partners look for — the actual substrate, not the narrative around it.

qai-system — spec-sheet v2 — QAI-SPEC-001 · hinch-2024-quaternion
Compute & Execution Layer
algebraQuaternion · q = w+xi+yj+zk
forward_passHamilton product (non-commutative)
inference_substrateS³ Manifold / Argus engine
activation_fnReLU · post-Hamilton product
model_frameworksOpen-weight · GGUF
param_count49,170 (published)
benchmark_accuracy98.04% · MNIST validation
benchmark_loss0.0831 · <2s train time
hardware_agnosticTrue · CPU · GPU · cluster
cloud_dependencyNone required
air_gap_nativeTrue
Transport & Security Stack
transport_layerML-KEM-768 · ML-DSA-65
key_exchangeFuture-resistant · post-quantum
edge_analysison-device · metadata-first
threat_responsegoverned agents · offense + defense
execution_modelZero-Trust AI Fabric
data_transmissionLocal · minimized exposure
deployment_targetsConsumer · edge · enterprise
data_localizationClient-owned
model_accessPublic + gated (Hugging Face)
quantizationGGUF · active R&D
research_since2014 · Derek Hinch
Live Shell · qctl v2.1 · read-only

Explore the
architecture.

A read-only window into the system architecture. Select a command to see how the stack is structured — every number traces to published benchmarks.

operator@qore:~ — qctl interactive
# Welcome to qctl. Select a command below. operator@qore:~$
One More Thing

Qore doesn't have to
work alone.

Turn on the Peer Mesh, and one Qore finds another — across the hall or across the world — to work as a single mind. Encrypted knowledge transfer. Signed delegation. Cryptographic provenance. Every link is post-quantum secured, and every peer is one you personally trust.

A private nervous system for your entire organization — with no cloud in the middle.

Explore the Peer Mesh
The Bottom Line · Genesis Mission Ready
Your intelligence.
Your hardware. Your rules.

The most powerful AI in the world should answer to one person: you. Qore runs on your hardware, keeps every byte in your custody, and proves every claim. Deployable today — no hyperscale dependency required.

Bring AI Home See the Platform
Our Story

The story behind
Qompute AI.

Built on a simple idea: artificial intelligence should be secure, efficient, and accessible. While much of the industry chased larger, more resource-hungry systems, founder Derek Hinch focused on a different challenge — delivering powerful AI through efficient architectures, trusted execution, and practical deployment. That work became the foundation for Qore. Today, under the leadership of Dale Zeller and the technical direction of Derek Hinch, Qompute AI builds technologies spanning artificial intelligence, information security, healthcare, digital forensics, and critical infrastructure.

2014
Research Begins
Derek Hinch begins exploring new approaches to computational efficiency, data security, and scalable intelligence systems.
2018
Core Breakthroughs
Early advances establish the foundation for secure data processing, high-efficiency architectures, and emerging AI capabilities.
2022
Architecture Defined
The technical framework evolves into Qore — a unified platform for deployable AI, secure infrastructure, and specialized systems.
2024
Technologies Expand
Development extends across artificial intelligence, secure communications, edge analysis, digital forensics, and healthcare applications.
2025
Deployment Accelerates
Qore continues expanding across security, healthcare, infrastructure, and investigative environments.
Leadership · The Team Behind Qompute AI

A world-class team
across every domain.

Bringing together technical innovation, information security, defense operations, and real-world experience — the founders and leaders building the future of secure, deployable intelligence.

DZ
Dale Zeller
Chief Executive Officer
Retired USAF Lieutenant Colonel with decades across intelligence operations, infrastructure resilience, and enterprise systems. Former J6 (CIO) for Joint Task Force Five, CIO of a California bank, Program Manager for the $2B Milstar Satellite Terminal Program, and Information Security Architect at Southern California Edison — where he authored the State of California's Information Security Plan. Leads Qompute AI's strategy, operations, and compliance.
Intelligence OpsInfrastructureLeadership
DH
Derek Hinch
Chief Technology Officer
AI architect and cybersecurity engineer with nearly three decades across U.S. Air Force Electronic Warfare R&D, hyperscale cloud security at Amazon, and Fortune 500 enterprise defense. Published author of the quaternion neural architecture powering the Argus engine — enabling mission-grade AI inference on field-deployable hardware at a fraction of conventional parameter counts.
Quaternion AIEdge IntelligenceSecurity Eng
WC
William Cronin
Chief Financial Officer
35+ years in financial services and corporate leadership. Founder of Madison Financial Services, with extensive public-company governance experience including 13+ years on an international manufacturer's board chairing its Finance and HR committees.
FinanceGovernanceStrategy
KK
Karen A. Kerames, CPA
Treasurer
CPA with 30+ years across finance, enterprise systems, and operational consulting, plus 22+ years of Oracle ERP expertise. International work spanning a dozen countries across manufacturing, pharmaceutical, banking, education, and government.
CPAOracle ERPSystems
JS
Josh Schwartz
Chief Marketing Officer
Leads brand, marketing, and go-to-market strategy across Qompute AI's portfolio of deployable intelligence, secure infrastructure, and specialized systems — shaping how the company's technologies reach the environments that matter most.
BrandGo-To-MarketGrowth
JV
Joey Victorino
VP, Infrastructure & Security
Senior cybersecurity executive with 10+ years on the front lines of incident response — nation-state intrusions, ransomware campaigns, APTs — across elite practices at a global cloud giant, a leading EDR firm, and a Fortune 100 tech consultancy. M.S. Cybersecurity. CISSP + 11 GIAC certifications (forensics, malware RE, detection engineering). Fluent English, Spanish, Russian. Directs infrastructure, security architecture, and post-quantum cryptographic research at Qompute AI.
CISSP + GIAC ×11Nation-State IRPost-Quantum Sec
DG
Desirae Gams
Chief of Staff
Drives organizational execution, cross-functional coordination, and operational cadence across Qompute AI — keeping leadership, engineering, and partnerships aligned as the company scales its technology portfolio.
OperationsCoordinationExecution
FF
Frank Falbo
Chief Operating Officer
Licensed professional engineer and attorney in Texas with 35+ years across complex industries. Began his career at Shell Oil before leading and scaling multiple ventures; managed a $900M+ annual drilling budget at Penn Virginia, helping position the company for a $1.7B merger offer. Leads Qompute AI's operational strategy and execution.
OperationsEngineeringLegal
TR
Tuck Rossmiller
Director of Public Relations
Leads public relations and external communications for Qompute AI, shaping the company's narrative across defense, technology, and investment audiences and managing media and stakeholder engagement.
CommunicationsMediaNarrative
We're Growing
Join Our
Mission
Help build the future of artificial intelligence, trusted systems, and secure computing.
Get in Touch
Careers · Join the Mission

Build the future of
sovereign AI.

We're a small, senior team building Qore — secure, deployable AI for the missions that can't use the cloud. If you want to work on hard problems that matter, alongside operators who've been inside the breach, we want to hear from you. Based in Las Vegas, NV · remote-friendly for the right people.

Senior AI / ML Engineer
Engineering · Las Vegas / Remote
  • On-device & edge inference
  • Agentic orchestration for Qore
  • Quantization & model optimization
Post-Quantum Cryptography Engineer
Security · Las Vegas / Remote
  • ML-KEM / ML-DSA implementation
  • Secure transport & key management
  • Immutable, signed audit systems
Security Research Engineer
Research · Remote
  • Offensive & defensive research
  • Detection & threat analysis
  • Feeds governed agent workflows
Solutions Architect — Air-Gapped & Federal
Field · Remote / Washington, DC
  • Sovereign & SCIF deployments
  • Compliance & control mapping
  • Customer technical leadership
Federal Capture & Partnerships Lead
Go-to-Market · Washington, DC / Remote
  • Government & defense engagement
  • RFI / RFP, SBIR / OTA pathways
  • Strategic partnerships
Platform / Full-Stack Engineer
Engineering · Las Vegas / Remote
  • Qore interfaces & governance UX
  • Audit, policy, and approval flows
  • Performance on constrained hardware
Executive Assistant
Operations · Las Vegas
  • Support founders & leadership
  • Scheduling, travel & communications
  • Discretion in a secure environment
Apply / Get in Touch

Don't see your role? Tell us what you'd build — careers@qompute.ai

Investors & Partners · QAI-CAP-001

Built to survive
diligence.

We raise the way we build. Every claim on this page traces to primary evidence. Every benchmark carries its scope. Every capability is labeled shipped or roadmap. What you see is what survives the data room — because the first team to pressure-test our numbers is us.

01
A Defensible Moat
Published, Benchmarked Research
The quaternion neural architecture behind our efficiency work is published (Hinch-2024-QNN): 98.04% test accuracy on MNIST at 49,170 parameters. A real, reproducible parameter-efficiency result — scoped to MNIST validation. Extending it to production architectures is active R&D, and we label it exactly that way.
02
Operators, Not Tourists
A Team You Can Diligence
A published quaternion-NN author as CTO. A retired USAF Lt. Colonel as CEO. Security leadership drawn from USAF / intelligence and Fortune-100 incident response. A CPA and a seasoned CFO/COO. Verifiable credentials and named backgrounds — available for reference, not just logos.
03
A Mandate, Not a Fad
Aligned to Data Sovereignty
Sovereign, local, air-gap-capable AI with post-quantum-oriented transport speaks directly to the data-custody problem the Genesis Mission names. To be precise: this is alignment of thesis and market — not a government award, endorsement, affiliation, or selection.
04
The Data Room Is Ready
A Substantiation Index
Every quantitative and capability claim we make is maintained with its evidence, scope, and caveats — and shared with qualified parties under NDA. Where a number needs earning, we name the benchmark, pilot, or hire that earns it. Conviction that holds up under scrutiny.
Request the Investor Briefing Explore the Platform

For accredited investors and institutional partners. This website is informational only and is not an offer to sell or a solicitation to buy any security; any offering is made solely through definitive documents via counsel. See disclosures below.

Contact · QAI-INV-001

Get in touch
with Qompute AI.

Interested in our technologies, partnerships, or capabilities? Send us a message and a member of our team will respond. Whether you represent an enterprise, a government institution, a research organization, or an investment partner — we want to understand what you're building.

Responsive communication
Trusted engagement
Enterprise support
QAI-INV-001 · Start the Conversation · Confidential
Helps us route your message to the right team