Articles
In-depth research, comprehensive guides, and long-form thoughts on modern technology.
Build Kubernetes Pod Networking by Hand to Understand CNI
A hands-on walkthrough wires up Kubernetes-style pod networking with raw Linux tools, showing exactly what a CNI plugin automates.
How Six Invisible Heading Tags Cost One Site AI Citations
A homepage scored 65 out of 100 on AI extractability until six mislabeled card-title headings were quietly demoted, no words removed.
GitLab's New Trigger Automates Work Item Assignment
GitLab's Work Item Created trigger fires an AI flow the moment an issue appears, automatically assigning it based on real team workload.
How Cursor and GitLab Team Up to Modernize Legacy Java
Cursor handles focused Java 8 to 21 fixes while GitLab's MCP server, CI, and review flows certify each change before it ships.
Copilot vs Raw API: What You're Really Paying For
Choosing between GitHub Copilot and raw model API access comes down to what layer of software work you actually need to own.
Karpathy Retrains a 1989 Neural Net With Modern Tricks
Andrej Karpathy reproduces Yann LeCun's 1989 digit-recognition paper, then applies 33 years of R&D to see how much error it can shed.
Karpathy's microgpt Packs GPT Training Into 200 Lines
Andrej Karpathy's microgpt trains and runs a GPT in 200 dependency-free lines of Python, distilling a decade of LLM research into one file.
Spotify Encodes Domain Expertise in Data Assistant Context Layer
Spotify's data assistant Vedder uses a curated context layer of domain expertise to provide reliable, trustworthy data insights.
Spotify's Data Assistant Uses Context Layer for Reliable Insights
Spotify built an AI data assistant called Vedder that uses a curated context layer to provide reliable data insights in seconds.
Netflix Builds Real-Time Service Topology at Scale
Netflix built a real-time service dependency map processing millions of flow records per second to help engineers troubleshoot faster.
Netflix Builds In-House LLM Serving Platform with vLLM and Triton
Netflix built an in-house LLM serving platform using vLLM and NVIDIA Triton to run models directly in their production environment.
Cursor Report Reveals AI Coding Habits of Top Developers
Cursor's 2026 developer habits report shows the top 1% of users generate 30-40K lines of code per week with AI assistance.