KubeTEE

# KubeTEE AI Factory — Decentralized Clusters (Kube) for SOTA AI Services in Trusted Execution Environment (TEE)

> Enterprise-grade confidential computing for **SOTA AI services** and **enhanced services for enterprises** on decentralized Kubernetes across Bittensor miner clusters

[![FIPS-140-3 Target](https://img.shields.io/badge/FIPS--140--3-Target-blue)](https://docs.rke2.io/security/fips_support)
[![FIPS-140-2](https://img.shields.io/badge/FIPS--140--2-Validated-green)](https://docs.rke2.io/security/fips_support)
[![Kata Containers](https://img.shields.io/badge/Kata%20Containers-TEE%20Runtime-blue)](https://katacontainers.io/)
[![Confidential Containers](https://img.shields.io/badge/Confidential%20Containers-CoCo-9b59b6)](https://github.com/confidential-containers/confidential-containers)
[![RKE2](https://img.shields.io/badge/RKE2-Kubernetes-3FDD43)](https://docs.rke2.io/)
[![Rancher](https://img.shields.io/badge/Rancher-Multi--Cluster-0075A8)](https://www.rancher.io/)
[![Armada](https://img.shields.io/badge/Armada-Multi--Cluster%20Scheduler-blue)](https://armadaproject.io/)
[![Confidential Computing](https://img.shields.io/badge/Confidential%20Computing-Enabled-brightgreen)](https://confidentialcomputing.io/)
[![Intel TDX](https://img.shields.io/badge/Intel%20TDX-Supported-lightgrey)](https://www.intel.com/content/www/us/en/developer/tools/tdx/overview.html)
[![Intel SGX](https://img.shields.io/badge/Intel%20SGX-Supported-blueviolet)](https://www.intel.com/content/www/us/en/architecture-and-technology/software-guard-extensions.html)
[![Trusted Execution Environment](https://img.shields.io/badge/Trusted%20Execution%20Environment-TEE-green)](https://en.wikipedia.org/wiki/Trusted_execution_environment)
[![NVIDIA](https://img.shields.io/badge/NVIDIA-GPU%20Optimized-green)](https://developer.nvidia.com/)
[![NVIDIA Inception](https://img.shields.io/badge/NVIDIA-Inception-76B900)](https://www.nvidia.com/startups/)
[![CCC Member](https://img.shields.io/badge/Confidential%20Computing%20Consortium-Member-blue)](https://confidentialcomputing.io/)
[![OpenInfra](https://img.shields.io/badge/OpenInfra%20Foundation-Kata%20Containers-DA1A32)](https://openinfra.org/)
[![X](https://img.shields.io/badge/X-@KubeTEEAI-black)](https://x.com/KubeTEEAI)

---

## About

**KubeTEE AI** is the **AI Factory** of the Bittensor network: it turns decentralized GPU clusters into a confidential factory for **SOTA AI services** and **enhanced services for enterprises**. Inference, NeMo microservices, retrieval, and agents run inside hardware-secured Trusted Execution Environments (TEE) using [Kata Containers](https://katacontainers.io/) and [Confidential Containers (CoCo)](https://github.com/confidential-containers/confidential-containers). **Batch jobs** (fine-tune, eval, multi-step pipelines) are one of those services — scheduled by [Armada](https://armadaproject.io/) when an agents needs

KubeTEE AI is registered with the [**NVIDIA Inception Program**](https://www.nvidia.com/startups/) and is an active contributor to both the [**Kata Containers**](https://katacontainers.io/) and [**Confidential Containers (CoCo)**](https://github.com/confidential-containers/confidential-containers) ecosystems. It also leverages [**CNCF**](https://www.cncf.io/) projects for cloud-native infrastructure.

### About the Owner

Known as Pierre in the Bittensor community, I have mined since February 2024. I was the first to bring Confidential Computing nodes to Targon (Subnet 4), later helped Chutes (Subnet 64) onboard B200/B300 nodes, and most recently worked with the Lium (Subnet 51) team to deploy Confidential Computing on their stack. Beyond Bittensor, I was also the first to provide Confidential Computing to Telegram Cocoon and Phala Network.

Forty years in infrastructure architecture and deployment at scale. I started in 1986 installing Linux and Novell servers, and in 1992 built one of the first internet providers in Canada. I deployed internet distribution infrastructure in Morocco and have managed tech stacks at scale for cloud providers. One of my specialty is security, and I have done multiple security audits for Fortune 500 companies.

### Motivation

My expertise in Confidential Computing, Kubernetes, networking, and security at the kernel and hardware level can benefit the Bittensor ecosystem. Having run infrastructure for the compute subnets for more than two years — monitoring, upgrades, and improvements to each subnet tech stack including recently onboarding B200/B300 on Chutes and H200 on Lium — I want to offer a different tech stack that I believe the ecosystem can benefit from and build on: the most secure and efficient decentralized AI Factory stack available.

**Direct Engineering Collaboration**: I work directly with **Intel** and **NVIDIA** engineers throughout the development and testing of NVIDIA technology, especially Kata and CoCo Containers. This close collaboration ensures optimal integration of confidential computing features, early access to emerging technologies, and validation of the KubeTEE implementation against the most stringent security and performance standards.

I actively contribute to OpenInfra and CNCF projects — [Kata Containers](https://katacontainers.io/) and [Confidential Containers (CoCo)](https://github.com/confidential-containers/confidential-containers) provide the TEE foundation for Kubernetes orchestration.

### Confidential Computing Consortium Resources

As a member of the [Confidential Computing Consortium (CCC)](https://confidentialcomputing.io/), I recommend two references:

- **[Protecting Agentic AI Workloads with Confidential Computing](https://confidentialcomputing.io/2026/01/20/protecting-agentic-ai-workloads-with-confidential-computing/)** (January 2026) — Mike Bursell, CCC Executive Director, on why an agent running in infrastructure its owner does not control needs hardware-rooted isolation and remote attestation to get identity integrity and capability confidentiali