Blog
Writing on GPUs, LLMs, MLOps, Kubernetes — and mindset · 3517 posts
#2026-03 765#english 592#culture 264#deep-dive 254#kubernetes 247#career 229#ai 216#llm 208#devops 193#2026-04 146#security 141#database 114#observability 113#communication 109#history 107#architecture 100#productivity 96#finance 88#economy 84#mindset 81#psychology 80#ai-papers 79#food 78#it 78#travel 78#deep-learning 77#japanese 77#networking 77#performance 72#business-travel 70#linux 70#gpu 69#ai-agent 66#cs-fundamentals 63#postgresql 60#rag 58#self-improvement 55#learning 53#mlops 53#ai-platform 51
Word and PowerPoint Shortcuts — Documents Are Handled by Style, Slides by Object
Word and PowerPoint use the same ribbon, but the muscle memory they need is completely different, because Word handles flowing text and styles while PowerPoint handles objects on a canvas. This post splits the two apps i
2026-08-02 · 22 min read #word#powerpoint#shortcuts#office#productivityNovels Worth the Time — 30 Books and the Rule of the First 50 Pages
Novel recommendation lists are everywhere, but they tend to leave out exactly what you need to know: how long the book actually is, how hard it is, and whether it's the kind of book you should push through when the first
2026-08-02 · 22 min read #storytelling#novels#reading#world-literature#cultureHow These Models Were Actually Built — Dissecting the 2026 Open-Weight Pipeline
Reading the cards and technical reports of the open-weight models sitting near the top of Hugging Face as of August 2026, this post lays out the production pipeline in order, from data collection to quantized deployment.
2026-08-02 · 17 min read #llm#pretraining#moe#post-training#quantizationDashboards That Get Read and Alerts Worth Paging — Defining Questions, Variables, SLOs, and Alert Fatigue
A dashboard is not useful for looking pretty — it earns its keep by answering a fixed set of questions in order. This post covers how to write down the questions a panel should answer before building it, and how to struc
2026-08-02 · 15 min read #observability#grafana#alerting#slo#dashboardsDesigning Prometheus Metrics That Answer Questions — Choosing Types, Cardinality Budgets, and the Traps in rate and Quantiles
More metrics is not better — a metric earns its keep only by answering a question. This post starts by laying out which questions counters, gauges, and histograms each answer, and which calculations become impossible in
2026-08-02 · 17 min read #observability#prometheus#promql#metrics#cardinalityWhere to Start with Classic Films — 25 Essential Movies and the Order to Watch Them
Lists of classic films are everywhere, but almost none tell you what order to watch them in. This piece sorts 25 essential films not by ranking but by the specific problem a beginner runs into: when black-and-white feels
2026-08-02 · 18 min read #storytelling#film#classics#world-cinema#cultureRunning a GPU Cluster with Slurm — Knowing Why a Job Will Not Run Matters More Than Submitting
Everything you need to actually use Slurm on a GPU cluster in practice. Sets up the coordinate system of partition, QoS, and account first, then covers how to request GPUs, CPUs, and memory in an sbatch script and the bi
2026-08-02 · 14 min read #mlops#slurm#hpc#gpu-cluster#distributed-trainingA Map of the LLM Training Stack in 2026 — What Each Layer Does For You, and What It Hides
Sorts LLM training frameworks into three layers and maps out their lineage. The bottom layer is the execution engines — PyTorch distributed, DeepSpeed, Megatron-Core. The middle layer is training loops like torchtitan an
2026-08-02 · 13 min read #mlops#llm-training#pytorch#trl#frameworkWhat's Trending on Hugging Face Right Now — A Map of August 2026
As of August 2, 2026, I went through the Hugging Face trending list directly and organized models that are actually usable by purpose. In order — general LLMs, coding, embeddings and rerankers, vision, speech, image/vide
2026-08-02 · 18 min read #llm#huggingface#open-weights#model-selection#quantizationAMD vs. NVIDIA: What Actually Differs — Why the Stack Is the Problem, Not the Hardware
Breaks down the difference between AMD and NVIDIA GPUs across four layers — hardware architecture, software stack, porting path, and ecosystem maturity — without taking sides. Covers the mapping from SM to CU and from Te
2026-08-02 · 16 min read #amd#rocm#hip#nvidia#cudaIs Comfort From a Machine Real? — Why We Need to Change the Question
People really do feel better after talking to an AI. Dismissing that as an illusion isn't just rude — it's inaccurate. This piece looks at why responsiveness feels like being cared for, and digs into what chatbot-based m
2026-08-02 · 14 min read #humanities#ai#psychology#loneliness#relationshipsTool or Counterpart — What Can Still Be Said While Leaving the Consciousness Question Open
The first question people reach for in front of AI is "does this thing have consciousness." It is the hardest question, and probably not the most useful one. This piece sets that question down beside us without closing i
2026-08-02 · 13 min read #humanities#ai#philosophy#cognition#technologyMaking vLLM Fast — Configuration, Internals, and Where to Actually Touch the Code
A step-by-step walk through improving vLLM performance, starting from everything you can fix without touching code. Covers batching-related arguments, prefix caching, chunked prefill, and quantization choice first, then
2026-08-02 · 21 min read #vllm#llm-inference#paged-attention#benchmark#schedulerWhat We Lose by Delegating — Automation Doesn't Erode Skill Evenly
Nobody feels guilty using a calculator, but most people feel a little uneasy after sending off a document AI drafted for them. That asymmetry is the question behind this piece. It sorts out the territory where offloading
2026-08-02 · 14 min read #humanities#ai#cognition#automation#skillGPU Compiler and Framework Landscape — One Problem, Turning a Graph into a Kernel, a Different Answer at Every Layer
This post puts NVCC and PTX, LLVM, MLIR, Triton, torch.compile, XLA, IREE, and TVM on one map. Different names, different owners, but they all solve the same problem: turning a computation graph into an executable kernel
2026-08-02 · 21 min read #gpu#compiler#mlir#triton#pytorchWhat It Really Means to Hand-Tune a GPU Kernel — Making One Transpose Kernel 5x Faster
Starting from threads, warps, and the memory hierarchy, this post covers what it actually means to hand-modify a GPU kernel. It explains why occupancy is a symptom rather than a goal, and why most kernels are bound by me
2026-08-02 · 19 min read #cuda#gpu-kernel#nsight-compute#memory-bandwidth#performanceThe Math You Need for Robotics, in Order: And What You Can Safely Put Off
An answer to how far you actually need to take your math to build a robot arm. Organized into six branches in order: linear algebra, trigonometry and rotation representations, calculus and multivariable methods, differen
2026-08-02 · 25 min read #robotics#math#electronics#linear-algebra#controlWhat a Robot Arm Is Made Of: Links, Joints, and What Actually Drives Them
Buy six servos to build a robot arm and it almost always collapses under its own weight. This post starts from the precise definitions of link, joint, and end effector, explains what degrees of freedom actually count and
2026-08-02 · 29 min read #robotics#hardware#electronics#arduino#mathThree Layers of Writing a Kernel — Comparing CUDA C++, Triton, and CUTLASS on the Same Problem
Compares what changes when you approach the same GPU kernel by hand in CUDA C++, tile-by-tile in Python with Triton, or assembled from templates in CUTLASS and CuTe. We actually write a row-wise softmax in both CUDA C++
2026-08-02 · 18 min read #triton#cuda#cutlass#gpu-kernel#compilerKorean Film and Drama Roundup 2025-2026 — The Year Theaters Came Back, the Year OTT Got Reshuffled
A roundup of 18 Korean films and dramas that actually became talked-about from 2025 through the first half of 2026. Covers The King's Warden, which drew 16.91 million admissions, Park Chan-wook's No Other Choice, which p
2026-08-02 · 17 min read #culture#korean-film#k-drama#movie-recommendation#netflix