AMEVA Open-Source Foundation (AOSF)
Democratizing On-Device AI & Autonomous Systems Without Cloud Egress Dependency
AMEVA Open-Source Foundation (AOSF) operates as an independent technical public entity. Personal career records, contact info, and private portfolio of Founder Eunho Kim (@uno-km) are maintained strictly separately in the Personal Digital CV.
AOSF is an independent non-profit open-source foundation governed by transparent meritocracy. We foster sovereign on-device AI, autonomous multi-agent orchestration, and WebGPU autograd engines that run 100% locally on user hardware with zero cloud server costs.
Tier 1: Flagship Top-Level Projects (TLP)
Production-Ready · PyPI & npm Verified · Complete Software & Runtimes
AMEVA Workstation (Web/Desktop)
World's first 100% on-device WebGPU local AI (Qwen2.5 0.5B/1.5B/7B), 3-second MapReduce PDF/DOCX doc reader & editor, in-app video cutting, 1-sec AI background removal, and audio silence auto-cut.
Termux-Playwright
Non-root Chromium browser automation on Android Termux (ARM64). Direct CDP control for scraping and automated web pipelines.
Termux-Diffusion
Native on-device Stable Diffusion runtime for Android ARM64. Generates images locally without cloud servers or root permissions.
Termux-STT
Unified on-device Speech-to-Text framework combining Whisper.cpp, Vosk, and Sherpa-ONNX with pure Python speaker diarization.
Termux-Train
Ultra-lightweight on-device Tensor & DAG Autograd deep learning framework with SafeTensors zero-copy and LoRA fine-tuning.
AMEVA-Forge
Client-compute offloaded architecture Browser-Native WebGPU Autograd Engine with PyTorch compatibility, WGSL compute shaders, and zero memory waste buffer pool.
Official Released Packages & Ecosystem Matrix
| Project | Package (PyPI / npm) | Runtime Environment | Primary Capability | Status |
|---|---|---|---|---|
| AMEVA Workstation | Web / Desktop |
WebGPU, Next.js, WASM | 100% On-Device Local AI & Productivity Suite | Official Live App |
| Termux-Playwright | pip: termux-playwrightnpm: termux-playwright |
Android ARM64 Termux | Non-root Chromium Browser Automation & CDP | PyPI + npm Released |
| Termux-Diffusion | pip: termux-diffusionnpm: termux-diffusion |
Android ARM64 Termux | On-Device Stable Diffusion Image Generation | PyPI + npm Released |
| Termux-STT | pip: termux-sttnpm: termux-stt |
Android ARM64 Termux | Unified Whisper/Vosk STT & Pure Python Diarization | PyPI + npm Released |
| Termux-Train | pip: termux-train |
Android ARM64 Termux | Bionic Native Tensor, DAG Autograd & LoRA | PyPI Released |
| AMEVA-Forge | pip: ameva |
WebGPU / Pyodide WASM | In-Browser WebGPU Deep Learning & Training | PyPI Released |
Official Charter of the AMEVA Foundation
Adopted August 2026
"To liberate artificial intelligence and compute autonomy from centralized cloud enclosures by engineering high-performance, edge-first open-source software that operates entirely on sovereign end-user hardware."
Article I. Name and Core Purpose
1.1 The organization is formally established as the AMEVA Open-Source Foundation (AOSF).
1.2 The Foundation is governed by the following immutable core principles:
- Edge-First & Serverless Edge Architecture: All AI inference and automation pipelines are engineered to execute on end-user hardware (smartphones, browsers, edge boards) without mandatory cloud server dependencies.
- 100% Data Sovereignty: User prompts, inputs, and inference results remain strictly inside client-side execution boundaries without telemetry leakage.
- Public Good & Open Standards: Foundation deliverables are permanently licensed as open-source public goods free from commercial enclosure.
- Strict Separation of Concerns: Foundation institutional assets and governance are operated strictly independently from Founder Eunho Kim (@uno-km)'s private career track records.
Article II. Technical Scope
| Domain | Core Technologies | Primary Objectives |
|---|---|---|
| Mobile & Edge Runtimes | Android Bionic ARM64, Termux, NEON SIMD | Non-root browser automation (termux-playwright), on-device Stable Diffusion (termux-diffusion), and tensor autograd (termux-train). |
| Browser & WebGPU AI | WebGPU, WGSL, Pyodide WASM | Browser-native autograd engine (AMEVA-Forge) and 100% isolated sandboxes (MCP-Wasm-Toolkit). |
| Autonomous Systems | Docker, SRE Mesh, Multi-CLI | Multi-agent orchestration (Agent Orchestra), social simulation (Dead Internet Theatre), and resilient retry-backed telemetry (Data Harvester). |
Article III. Licensing & Intellectual Property
All canonical deliverables, libraries, runtimes, and documentation produced under AOSF are licensed under the Apache License, Version 2.0 or the MIT License.
Article IV. Disclaimer of Liability
Deliverables are provided "AS IS", without warranty of any kind, express or implied.
Governance & Meritocracy Guidelines
AOSF Operating Regulations & Meritocratic Bylaws
The AMEVA Open-Source Foundation operates strictly on meritocracy: influence, committer rights, and Project Management Committee (PMC) seats are earned exclusively through sustained, high-quality technical and architectural contributions.
1. Roles and Responsibilities
Community roles are categorized into Users, Contributors, Committers, and PMC Members with clear, transparent promotion criteria.
2. Decision-Making & Voting Protocols
Consensus building is prioritized, followed by formal 72-hour lazy consensus votes where required for major architectural changes.
3. Project Lifecycle
Projects progress through Sandbox -> Incubation -> Top-Level Project (TLP) based on code maturity and multi-maintainer diversity.
4. Conflict Resolution & Code of Conduct
All participants adhere to the Contributor Covenant Code of Conduct, ensuring a professional, harassment-free environment.
Incubation Policy & Lifecycle
Standard Process for Onboarding New Sub-Projects
The AOSF Incubation Policy governs the ingestion, mentorship, and graduation of emerging edge-first AI projects into official Tier 1 Top-Level Projects.
Stage 1: Proposal & Sponsor
A proposal outlining the technical scope, zero-cloud architecture, and initial maintainers is submitted for PMC review.
Stage 2: Mentored Incubation
The project establishes transparent issue tracking, CI/CD, documentation, and releases on PyPI/npm under AOSF guidelines.
Stage 3: Graduation to Tier 1 TLP
Upon demonstrating production stability, complete unit test suites, and meritocratic governance, the project graduates to TLP.
Graduation Requirements
Projects must exhibit 100% data sovereignty, edge-first execution, clean Apache 2.0/MIT licensing, and comprehensive API documentation.
Sponsorship & Institutional Support
Funding Public Goods & Sovereign Edge AI Infrastructure
AOSF is sustained by direct sponsorships, corporate grants, and compute hardware donations dedicated exclusively to open research and testbed infrastructure.
Hardware & Compute Grants
Testing devices (ARM64 devboards, smartphones, GPUs) are used directly in CI/CD pipelines to benchmark real hardware performance.
Ecosystem Sustainer Tier
Organizations providing annual financial grants receive prominent recognition across Foundation portals and documentation feeds.
Individual Backers
Individual developers supporting via GitHub Sponsors or OpenCollective help maintain domain infrastructure and package registries.
Transparency of Funds
All financial and hardware contributions are accounted for with public quarterly transparency reports.