Attention from Scratch: Soft Attention to Multi-Head
A step-by-step walk through attention: soft attention, weight-free self-attention, trainable QKV, causal masking, dropout, and multi-head. Embeddings are assumed to be already in hand.
Posts
Practical notes on AI/ML, data science, and systems.
Fighting Scraping Farms on a Reading Site
Field notes from fighting EPUB-scraping farms on BetterReads — good bots vs bad bots, why fingerprints fade, and how observability can cost more than the attack.
When CleanMyMac Isn't Enough: Building My Own Disk Cleaner
CleanMyMac missed my real disk hogs — a 175GB .next cache and a 60GB Cursor DB. I built a native macOS cleaner and reclaimed 500+ GB.
Image Processing with R (imager): Pixels, Filters & Examples
Learn image processing with R and the imager package — load images, inspect pixels/intensity, apply filters, and follow practical visual examples.
使用 R 处理图像
使用 R 与 imager 进行数字图像处理：像素、灰度、滤波与基础视觉操作。
Diving into Vector Database
How vector databases store embeddings and run similarity search — ANN indexes, ChromaDB/SQLite, and when to use them for semantic search and RAG.
Privacy-Aware RAG Based Knowledge Base
Build a privacy-aware local RAG knowledge base with agent support — no cloud LLM calls required.
隐私保护的RAG知识库
构建注重隐私与成本的本地 RAG 知识库，并加入智能体支持以提升检索与回答质量。
Text Cluster Analysis: Grouping Documents by Similarity
Text Cluster Analysis: Grouping Documents by Similarity
文本聚类分析：基于相似性的文档分组
使用 R 进行文本聚类分析实战：通过预处理、TF-IDF、余弦距离与 K-means/层次聚类，按内容相似性自动分组文档，含 20 Newsgroups 示例、PCA 可视化与词云主题探索。
Sentiment Analysis: Decoding Emotions in Text
Sentiment Analysis: Decoding Emotions in Text
情感分析：解码文本中的情感
情感分析：解码文本中的情感
Topic Modeling with LDA: Uncovering Hidden Themes in Text - Father's Day Edition
Topic Modeling with LDA: Uncovering Hidden Themes in Text
使用LDA进行主题建模：发现文本中的隐藏主题 - 父亲节特别版
使用LDA进行主题建模：发现文本中的隐藏主题
Unstructured Text Mining in R: A Beginner's Guide
Unstructured Text Mining in R: A Beginner's Guide
Text Exploration and Preprocessing in R: A Beginner's Guide
Text Exploration and Preprocessing in R: A Beginner's Guide
R语言文本探索与预处理：入门指南
R语言文本探索与预处理：入门指南
R语言非结构化文本挖掘入门指南
R语言非结构化文本挖掘入门指南
Key Techniques in Data Preparation for Machine Learning
Practical data preparation for machine learning: imputation, scaling, encoding, SVD intuition, and feature hygiene before you train supervised models.
机器学习数据准备的关键技术
机器学习数据准备实战：缺失值处理、特征缩放、编码、SVD 直觉，以及训练监督模型前的特征卫生检查。
使用R进行数字信号处理：婴儿哭声分析深度解析
用 R 做数字音频信号处理实战：以婴儿哭声音频为例，讲解 tuneR、seewave 的频谱与声谱图分析，从傅里叶变换到可视化的 DSP 流程，适用于语音识别、生物声学与医学诊断场景。
Hugging Face MCP with OpenAI - Zero Code Setup
Zero-code setup for Hugging Face MCP with OpenAI so agents can access models, datasets, and Spaces.
Hugging Face MCP 与 OpenAI 集成 - 零代码配置指南
零代码配置 Hugging Face MCP 服务器与 OpenAI，让 AI 智能体访问模型库、数据集与 Spaces。
Digital Signal Processing in R: A Deep Dive into Baby Cry Analysis
Analyzing Audio Signals in R: A Deep Dive into Baby Cry Analysis
Recurrent Neural Network (RNN): From Theory to Translation
Recurrent Neural Network (RNN): From Theory to Translation
循环神经网络(RNN)：从理论到翻译
循环神经网络(RNN)：从理论到实践
向量数据库简介
向量数据库如何存储嵌入并做相似性检索——ANN 索引、ChromaDB/SQLite，以及在语义搜索与 RAG 中的适用场景。
C语言变量存储与指针：基础篇
理解C语言中的变量存储类别和指针基础
C语言进程间通信：基础篇
使用匿名管道进行进程间通信
Variable Storage and Pointers in C: Back to Basics
Variable Storage and Pointers in C: Back to Basics
Inter Process Communication in C: Back to Basics
Interprocess communication (IPC) with anonymous pipes.
图卷积网络：从理论到实践
图卷积网络：从理论到实践
The Glorious Graphic Convolutional Network - From Theory to Practice
The Glorious Graphic Convolutional Network - From Theory to Practice
Classification and Logistic Regression
Classification with logistic regression: parametric vs non-parametric models, the sigmoid, decision boundaries, and a clear path from linear regression intuition.
分类与逻辑回归
用逻辑回归做分类：参数模型与非参数模型、Sigmoid、决策边界，以及从线性回归直觉过渡到分类问题的清晰路径。
Linear Optimization in Finance: Portfolio Allocation and Solvers - Part 2
Linear Optimization in Finance: Portfolio Allocation and Solvers - Part 2
金融中的线性优化：投资组合分配与求解器 - Part 2
用 Python、PuLP 与开源 CBC 求解器实现金融线性优化：在预算与风险等约束下最大化收益或最小化波动，完成投资组合分配、建模与实战求解，并衔接 Part 1 的线性金融定价基础。
Python in Finance - Part 1
Python in Finance - Part 1
Python 在金融中的应用- Part 1
Python 在金融中的应用- Part 1
From Linear Algebra to Linear Regression - the ML Perspective
From Linear Algebra to Linear Regression - the ML Perspective
从线性代数到线性回归——机器学习视角
从线性代数到线性回归——机器学习视角
Succinct Guide Guide on Text Classification with fastai - Fastai Part 5
Succinct Guide Guide on Text Classification with fastai - Fastai Part 5
使用 fastai 进行文本分类的简明指南 - Fastai Part 5
使用 fastai 进行文本分类的简明指南 - Fastai Part 5
Chinese Natural Language Processing with fastai - Fastai Part 4
Chinese Natural Language Processing with fastai - Fastai Part 4
中文NLP with fastai - Fastai Part 4
中文NLP with fastai - Fastai Part 4
Deep Dive into MNIST Dataset - Fastai Part 3
Deep Dive into MNIST Dataset - Fastai Part 3
深入探究 MNIST 数据集 - Fastai 第三部分
深入探究 MNIST 数据集 - Fastai 第三部分
Tschick - Eine Rezension
Rezension zu Wolfgang Herrndorfs Roman Tschick — eine Sommerreise zweier Teenager im geklauten Lada.
The Updated Guide to Image Classification - Fastai Part 2
The Updated Guide to Image Classification - Fastai Part 2
图像分类最新指南 - Fastai 第二部分
图像分类最新指南 - Fastai 第二部分
The Missing Bits - Easing into fastai and PyTorch - Part 1
The Missing Bits - Easing into fastai and PyTorch - Part 1
缺失的部分 - 轻松入门 fastai 和 PyTorch - 第一部分
缺失的部分 - 轻松入门 fastai 和 PyTorch - 第一部分
Diffusion Optimization - Optimizing Smart Energy Grid in a City
A practical example of diffusion optimization in smart energy grid optimization in a city
扩散优化 - 城市智能电网优化
扩散优化与随机梯度下降(SGD)优化方法的全面比较，以城市智能电网优化为例
Fractals Unveiled: From Mathematics to Interactive Art with TypeScript
Learn how to generate beautiful fractal images using TypeScript, understand the mathematics behind fractals, and explore their real-world applications in art, finance, and more.
分形艺术探秘：TypeScript 实现数学之美
学习如何使用 TypeScript 生成美丽的分形图像，了解分形背后的数学原理，探索其在艺术、金融等领域的实际应用。
梯度下降法原理解析：从数学基础到实现
一份适合初学者的梯度下降法全面指南，涵盖从单变量微积分到多元梯度的数学基础，配有清晰的解释和可视化示例，帮助读者深入理解这一机器学习中的核心优化算法。
Understanding Gradient Descent with a Sprinkle of Math
A beginner-friendly yet comprehensive guide to understanding gradient descent in machine learning, covering the mathematical foundations from single-variable calculus to multivariable gradients, with clear explanations and visual examples.
Understanding Perceptrons: The Building Block of Neural Networks (With Python Code)
Learn about perceptrons - the fundamental building blocks of neural networks. Includes practical Python implementation, visual explanations, and real-world applications like spam detection. Perfect for beginners in machine learning.
感知器：神经网络的基础构建块（含Python代码）
了解感知器 - 神经网络的基础构建块。包含实用的Python实现、可视化解释和垃圾邮件检测等实际应用。适合机器学习初学者。
Genetic Algorithm: From Darwin to Holland
Genetic algorithms from Darwin to Holland — theory, intuition, and an animated AlgoScope demo.
遗传算法：从达尔文到霍兰德
遗传算法从达尔文到霍兰德的演进讲解，配合 AlgoScope 动态可视化演示。
Exploring Optimization: A Guide to Iterative Local Search (ILS)
Learn Iterative Local Search (ILS): local search, perturbation, and acceptance to escape local optima—with Python code and a live AlgoScope visualization demo.
探索优化：迭代局部搜索（ILS）指南
迭代局部搜索（ILS）优化算法指南，含 AlgoScope 可视化演示。
可视化 A-Star 搜索
A* 寻路算法可视化讲解，附互动在线演示。
Visualizing A-Star Search
Visual walkthrough of A* pathfinding: how f(n)=g(n)+h(n) blends cost-so-far and heuristics vs Dijkstra, plus an interactive browser demo on AlgoScope.
使用R预测员工流失：实践指南（中文版）
使用 R 预测员工流失的实践指南：分类模型、特征分析与留任策略。
Predicting Employee Attrition with R: Logistic Regression Tutorial
Step-by-step R tutorial for employee attrition prediction — logistic regression, feature prep, and retention insights you can reuse.
To dropna() or not to dropna()
When to use pandas dropna() vs imputation for missing data in machine learning, with BRFSS examples.
A Return after a Long Departure
Returning to blogging after nine years away — reflections on Europe, career, and starting to write again.
A Proper Goodbye
Reflections on leaving Easy Taxi after 15 months of product and organization transformation in Brazil.
Taming Postgres on Heroku
How to diagnose and fix slow Postgres on Heroku as your app grows — indexes, cache, and practical tuning tips.
An Unbiased Yogic Guide to San Francisco
An unbiased guide to yoga studios and teachers in San Francisco, from Iyengar to Ashtanga.
On Agile
Thoughts on agile software development, rules programmers resist, and what actually helps teams ship.