全新开源版本
4 分•作者: BrainDAnderson•4 天前
我将发布三个大型软件系统，这些系统是我独自一人在过去几年里私下开发的，没有得到任何机构支持、企业赞助或正式的工程团队的帮助。
这些不是概念或模型。它们是真实的、可部署的系统。它们通过 Docker、Helm 或 Kubernetes 安装，可以成功启动并产生可观察的结果。它们目前正在云基础设施上运行。也就是说，应该将它们理解为未完成的基础，而不是经过打磨的产品。
整个生态系统总共大约有 150 万行代码。
平台
ASE — 自主软件工程系统 ASE 是一个用于代码创建、监控、评估和迭代自我改进的闭环平台。其目的是通过从高级任务中生成工件、监控结果、将更正反馈到流程中，并随着时间的推移不断完善自身，从而实现软件生命周期部分的自动化。它现在可以运行，但代理调整不一致，某些功能未完成，并且输出质量因配置而异。
VulcanAMI — Transformer / 神经符号混合 AI 平台 VulcanAMI 是一个基于混合架构构建的 AI 系统，该架构结合了基于 Transformer 的语言建模与结构化推理、编排和治理。目标是通过集成符号和系统级控制来解决纯统计模型的弱点。它可以部署和运行，但可靠的 Transformer 集成仍然是一个主要挑战。
FEMS — 有限庞大引擎 FEMS 是一个用于大规模场景探索、反事实分析和因果建模的实用多重宇宙模拟平台。它旨在应用通常仅限于研究环境的技术。它可以运行并产生结果，但其模型需要专家级的数学调整，不应被视为经过科学验证。
当前状态
所有这三个系统都是：
可部署的
可操作的
复杂的
未完成的
已知的局限性包括粗糙的用户体验、某些地方的文档不完整、与生产软件相比，正式测试有限、以可行性为导向的架构、安全加固不完善以及存在错误。
为什么现在发布
进一步的进展将受益于外部专业知识和观点。作为一名独立开发者，我缺乏充分完善如此大规模项目的资源。这次发布与资金、产品发布或机构计划无关。它仅仅是开放了已经存在、正在运行但尚未完成的工作。
关于我
我的名字是 Brian D. Anderson。我的主要职业是奇幻小说作家，而不是传统的软件工程师。我是自学成才的，在晚年开始学习软件系统，并且在消费级硬件上独立构建了这些平台。
这种背景会引来质疑。这也解释了这项工作的特点：雄心勃勃、打磨不均匀，并且由毅力而非正式流程驱动。构建这些系统是因为我希望它们存在。
这是什么——以及不是什么
这是：
一套可部署的基础
正在进行的独立工作的快照
一个探索、批判和改进的邀请
迄今为止已构建内容的记录
这不是：
一个成品套件
一个交钥匙解决方案
对突破性性能的声明
对支持或路线图的承诺
如果您探索代码，请假设某些部分过度设计，另一些部分开发不足，命名可能不一致，并且许多内部知识尚未外部化。如果任何部分有用或有趣，您可以在许可条款下自由地在其基础上进行构建。
这些系统存在。它们正在运行。它们尚未完成。如果它们对其他人有用，那就足够了。
— Brian D. Anderson
代码库：The Code Factory: https://github.com/musicmonk42/The_Code_Factory_Working_V2.git
VulcanAMI: https://github.com/musicmonk42/VulcanAMI_LLM.git
FEMS: https://github.com/musicmonk42/FEMS.git
查看原文
I am releasing three large software systems I developed privately over several years as a solo effort, without institutional backing, corporate sponsorship, or a formal engineering team.
These are not concepts or mockups. They are real, deployable systems. They install through Docker, Helm, or Kubernetes, start successfully, and produce observable results. They are currently running on cloud infrastructure. That said, they should be understood as unfinished foundations, not polished products.<p>Together, the ecosystem totals roughly 1.5 million lines of code.<p>The Platforms<p>ASE — Autonomous Software Engineering System ASE is a closed-loop platform for code creation, monitoring, evaluation, and iterative self-improvement. Its purpose is to automate parts of the software lifecycle by producing artifacts from high-level tasks, monitoring outcomes, feeding corrections back into the process, and refining itself over time. It runs today, but agent tuning is uneven, some features are incomplete, and output quality varies by configuration.<p>VulcanAMI — Transformer / Neuro-Symbolic Hybrid AI Platform VulcanAMI is an AI system built around a hybrid architecture combining transformer-based language modeling with structured reasoning, orchestration, and governance. The goal is to address weaknesses in purely statistical models by integrating symbolic and system-level controls. It deploys and operates, but reliable transformer integration remains a major challenge.<p>FEMS — Finite Enormity Engine FEMS is a practical multiverse simulation platform for large-scale scenario exploration, counterfactual analysis, and causal modeling. It aims to apply techniques often confined to research settings. It runs and produces results, but its models require expert mathematical tuning and should not be treated as scientifically validated.<p>Current Status<p>All three systems are:<p>Deployable<p>Operational<p>Complex<p>Incomplete<p>Known limitations include rough user experience, incomplete documentation in places, limited formal testing compared to production software, feasibility-driven architecture, incomplete security hardening, and the presence of bugs.<p>Why Release Now<p>Further progress would benefit from outside expertise and perspective. As a solo developer, I lack the resources to fully mature projects of this scale. This release is not tied to funding, a product launch, or an institutional program. It is simply the opening of work that exists, runs, and remains unfinished.<p>About Me<p>My name is Brian D. Anderson. My primary career has been as a fantasy author, not a traditional software engineer. I am self-taught, began learning software systems later in life, and built these platforms independently on consumer hardware.<p>That background will invite skepticism. It also explains the character of the work: ambitious in scope, uneven in polish, and driven by persistence rather than formal process. These systems were built because I wanted them to exist.<p>What This Is — and Is Not<p>This is:<p>A set of deployable foundations<p>A snapshot of ongoing independent work<p>An invitation to explore, critique, and improve<p>A record of what has been built so far<p>This is not:<p>A finished product suite<p>A turnkey solution<p>A claim of breakthrough performance<p>A promise of support or roadmap<p>If you explore the code, assume some parts are over-engineered, others under-developed, naming may be inconsistent, and much internal knowledge is not yet externalized. If any part is useful or interesting, you are free to build on it under the license terms.<p>The systems exist. They run. They are unfinished. If they prove useful to someone else, that is enough.<p>— Brian D. Anderson<p>Repos: The Code Factory: https://github.com/musicmonk42/The_Code_Factory_Working_V2.git<p>VulcanAMI: https://github.com/musicmonk42/VulcanAMI_LLM.git<p>FEMS: https://github.com/musicmonk42/FEMS.git