Maarten Grootendorst

Maarten Grootendorst 是一位荷兰数据科学家和 Google DeepMind 的人工智能工程师，他最出名的是创建了 BERTopic 和撰写了《Hands-On Large Language Models》一书。他形容自己是一位从心理学家转型的人工智能工程师，这种背景塑造了他以人为本的机器学习方法。

他的工作专注于通过高度可视化的指南和开源工具使人工智能易于访问。他为 Google 的 Gemma 模型系列做出了贡献，致力于优化工作，使大型语言模型在消费级硬件上运行得更快。他的教育内容面向广大开发者和研究人员。 [编辑]

Maarten Grootendorst 在荷兰长大，从小就对理解人类行为很感兴趣。他学习了心理学，这个领域教会了他人们如何思考和学习。但他很快意识到数据可以提供更深入的见解，于是他转到 Jheronimus Academy of Data Science 学习数据科学。

这条跨学科的道路使他与众不同。虽然他的大多数同学都来自计算机科学专业，但他带来了关于如何设计服务于人的 AI 系统的独特视角。在学习期间，他开始尝试自然语言处理，被教导机器理解文本的挑战所吸引。

他的早期项目包括尝试主题建模，这项技术后来成为了 BERTopic。他曾对 LDA 等旧方法的局限性感到沮丧，并开始开发自己的方法。这些融合了心理学和机器学习的 formative years 为他后来的突破奠定了基础。[编辑]

Grootendorst 的突破来自于 2022 年 BERTopic 的发布。该技术使用 BERT 嵌入来识别文本中的主题，并在 NLP 社区迅速获得关注。它在性能上超越了 LDA 等传统方法，并成为数据科学家们首选的工具。

2024 年，他与 O'Reilly Media 合作出版了 Hands-On Large Language Models。这本书教授了处理 LLM 的实用方法，并在 AI 教育界广受推荐。它巩固了他作为应用 AI 领域领先教育者的声誉。

他加入了 Google DeepMind，在那里他为 Gemma 系列开源模型做出了贡献。2026 年 5 月，他致力于通过多令牌预测草稿器加速 Gemma 4 的推理，速度提高了 3 倍。Gemma 4 12B 模型于 2026 年 6 月 3 日发布，仅需 16GB VRAM 即可运行，使其在笔记本电脑上也能轻松使用。

除了日常工作，他继续创作教育内容，包括在 Towards Data Science 上的教程以及在行业活动中的演讲。他的工作始终在复杂的 AI 研究与实际应用之间架起桥梁。[编辑]

Grootendorst 被视为一位平易近人、以教育为先的 AI 研究员。他将自己定位为连接复杂技术概念与实际理解的桥梁，以其标志性的视觉化、易于理解的解释风格而闻名。他的内容基调充满热情，专注于揭开 AI 的神秘面纱。

在数据科学界，他因其开源贡献（尤其是 BERTopic）以及清晰阐释复杂概念的能力而受到尊敬。他作为一名心理学家转型为 AI 工程师的职业身份，为他的品牌增添了以人为本的维度。[编辑]

Maarten Grootendorst commands a following defined not by celebrity worship but by deep professional gratitude. His audience consists of data scientists, researchers, and developers who rely on his open-source libraries like BERTopic and his visual guides to navigate the complexities of artificial intelligence. This is a community of practitioners who view him as a mentor rather than an idol, engaging with his work through GitHub contributions, Substack updates, and technical discussions rather than fan rituals or crowds. There is no formal name for this collective; they are simply the users who benefit from his mission to make AI accessible and human-centered. While some might imagine a bustling fanbase with inside jokes and organized campaigns, the reality is a quiet, dedicated network of professionals who appreciate his unique ability to bridge psychology and engineering. The relationship is built on the utility of his tools and the clarity of his explanations, creating a legacy of knowledge sharing that transcends typical entertainment fandoms. iFANN has evaluated Maarten Grootendorst within its Global Fame Engine framework. The assessment confirms that he currently does not hold a placement in the published rankings, including the Global Top 1,000 or any Top-100 scope list. He remains an unranked figure in the current global fame standings.[编辑]

Maarten Grootendorst operates under a personal motto found on his GitHub: using AI to make the world a slightly better place. This philosophy drives his transition from psychology to AI engineering, a pivot that informs his human-centered design approach. He distinguishes himself in the field by creating highly visual guides that demystify artificial intelligence concepts, setting him apart from educators who rely heavily on dense text. His book, Hands-On Large Language Models, was published by O'Reilly Media, cementing his status as a key voice in the industry. He maintains an active Google Scholar profile to track the citations of his academic contributions. While there are no folkloric legends or viral crowd incidents associated with him, his creation of the BERTopic library stands as a verified milestone that reshaped topic modeling standards. The publication of his O'Reilly title further solidified his role as a bridge between theoretical research and practical application for developers worldwide. He contributes significantly to Google's Gemma model family, working on optimizations that make large language models run faster on consumer hardware. His Substack newsletter, Exploring Language Models, serves as a primary hub for sharing these visual updates with a broad audience of practitioners.[编辑]

关于 Maarten Grootendorst 的家庭细节并未公开。他保护自己的个人生活隐私，将公众形象完全集中在他在人工智能和数据科学领域的专业工作上。目前没有关于他的父母、兄弟姐妹或家庭成员的公开信息。[编辑]

Grootendorst的专业网络包括在Google DeepMind从事Gemma模型系列工作的同事、BERTopic开源项目的贡献者、O'Reilly Media的编辑团队，以及Towards Data Science出版社区。[编辑]

关于Maarten Grootendorst的恋爱关系或婚姻状况，目前没有公开信息。他始终保持着严格的职业公众形象，截至2026年，他的感情状况仍然未知。[编辑]

As of 2026, Maarten Grootendorst's net worth is estimated at $2 Million per iFANN's editorial analysis. His primary income stems from his position at Google DeepMind, where senior AI research engineers typically command salaries between $150,000 and $400,000 annually alongside equity compensation. Additional revenue flows from royalties generated by Hands-On Large Language Models and fees from speaking engagements and consulting work. This financial standing has grown steadily since the release of BERTopic and his subsequent publishing success, mirroring his rising prominence within the artificial intelligence sector. While exact figures remain private, the combination of a top-tier technology salary and a thriving career as an author supports this valuation.[编辑]