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
Preparing to Become a Forward Deployed Engineer: A Map of the Software Knowledge That Matters
A Forward Deployed Engineer (FDE) is a role Palantir invented: an engineer who embeds with a customer and deploys, integrates, and configures the product inside the client's real environment. The role is surging again as
2026-07-11 · 6 min read #career#fde#software-engineering#palantir#solutions-engineeringProduction RAG Patterns — Why Naive RAG Fails and the Techniques That Actually Help
A demo RAG system takes half a day to build, but the place it quietly breaks in production is almost always retrieval, not generation. This reference walks through chunking, embeddings and hybrid search (BM25 + vector),
2026-07-11 · 9 min read #ai#rag#llm#retrieval#embeddingsThe Three Tribes of Automation — Zapier/Make/n8n, RPA, and 2026 Agentic Automation
When someone says "just automate this," people often reach for the wrong tool. That is because automation has three distinct tribes — workflow automation that connects APIs (Zapier/Make/n8n), RPA that operates screens li
2026-07-09 · 6 min read #automation#rpa#zapier#n8n#ai-agentsGPU Operator × KubeVirt Complete Guide — Components, Configuration, Versions, Partial MIG, and Manual MIG
The two pillars of GPU infrastructure on Kubernetes, all on one page. Covers the GPU Operator operands, ClusterPolicy configuration, and versioning scheme, plus a custom config that applies MIG to only some of the GPUs o
2026-07-09 · 9 min read #kubernetes#gpu#kubevirt#mig#nvidiaThe Complete Guide to LLM Training Data Preprocessing — From Web Crawls to Token Packing, with the Latest Papers
Good models come from good data, and good data comes from a preprocessing pipeline. This post walks through the entire pretraining data process step by step — web crawl collection → text extraction → language identificat
2026-07-09 · 7 min read #ai#llm#data-engineering#preprocessing#trainingThe 2026 Robotics Company Map — The Humanoid Showdown, VLA Models, and the Engineer's Way In
2026 is the inflection point where humanoid shipments jump about 7x year-over-year to a forecast of 50,000+ units. From Figure, valued around USD 39B with its in-house VLA model Helix, to Tesla Optimus Gen 3 entering mas
2026-07-09 · 7 min read #robotics#ai#vla#trends#careerTwo Paths to 3D — Reconstruction (NeRF & Gaussian Splatting) and Generation (TRELLIS & Hunyuan3D)
The phrase "make a 3D model" hides two completely different problems. Reconstruction brings a scene that actually exists back to life from a handful of photos, while generation conjures something that does not exist from
2026-07-09 · 9 min read #3d#ai#gaussian-splatting#nerf#computer-visionBuilding SSO with Keycloak — From Realm, Client, and Flows to the 2026 New Features
Instead of bolting a separate login onto each of 20 internal apps, SSO lets a single identity server stand in for all of them. This post makes sense of Keycloak — the open-source standard — through four core concepts (Re
2026-07-09 · 9 min read #keycloak#sso#oidc#security#devopsMulti-GPU, Multi-Node Training Platforms: The Complete Map — from the Framework Ecosystem to Hands-On Slurm and Kubeflow Guides
A one-page map of the full landscape of training models across multiple GPUs and multiple nodes. The AI library and framework ecosystem map (PyTorch, JAX, HuggingFace, DeepSpeed, Ray), when to pick which parallelization
2026-07-09 · 8 min read #ai#ml#distributed-training#slurm#kubeflowFrom Docker to Podman — the Complete Guide to Switching to a Daemonless Container Engine
Podman differs from Docker at the level of architectural philosophy: daemonless (fork-exec) operation and rootless by default. Internals (conmon, crun), config file locations and meanings, CDI setup for GPUs, the two way
2026-07-08 · 8 min read #docker#podman#containers#devops#linuxLLM Caching, Explained — Why Prompt Caching and Prefix Caches Save You Money
Why does a matching prompt prefix cut costs to a tenth? The answer lives in the transformer's KV cache. Because attention is causal, the Key/Value vectors of earlier tokens never change no matter what comes after — so th
2026-07-08 · 7 min read #ai#llm#caching#inference#performanceThe State of LLM Quantization — From GPTQ and AWQ to FP8, MXFP4, and KV-Cache Quantization
Quantization represents a model's numbers in fewer bits to cut memory and cost. Why quality survives fewer bits (outliers and scaling), how GPTQ and AWQ differ in approach, llama.cpp GGUF k-quants, QLoRA's NF4, the 2026
2026-07-08 · 7 min read #ai#llm#quantization#inference#optimizationThe Ownership Backlash and the AI Reality Check — This Week's HN & GeekNews Hot Topics
Two currents run through this week's Hacker News and GeekNews front pages: an 'ownership backlash' spreading from open-source maps to open hardware and replaceable batteries, and an 'AI reality check' driven by open-weig
2026-07-07 · 5 min read #trends#hackernews#geeknews#ai#opensourceThe Forward Deployed Engineer (FDE), Fully Dissected — the Hottest Role in AI Right Now
As the finding that 95% of enterprise GenAI pilots show no measurable impact makes clear, AI's bottleneck today is deployment, not models. The role built to close that gap is the FDE — an engineer embedded in the custome
2026-07-07 · 8 min read #career#fde#ai#palantir#rolesAI Model Development, Start to Finish — a Realistic Lifecycle from Data to Deployment
Model development starts not with pretraining but with a decision ladder — does prompting suffice, does RAG suffice, do you need fine-tuning? The eval-first principle of building your test set before your model, data qua
2026-07-07 · 8 min read #ai#ml#llm#mlops#trainingMastering the NVIDIA GPU Operator — From Install and Deployment to MIG Partitioning
The era of hand-configuring GPU nodes on Kubernetes is over. How the NVIDIA GPU Operator manages everything from the driver to the device plugin and monitoring via the operator pattern, with Helm install and verification
2026-07-07 · 8 min read #kubernetes#gpu#nvidia#mig#devopsHow to Use OpenRouter — 300+ LLMs Through One API
How to call 300+ models from 60+ providers through a single OpenAI-compatible endpoint with OpenRouter — from getting a key and your first curl request to SDK drop-in, routing, fallbacks, and streaming.
2026-07-06 · 8 min read #openrouter#llm#ai-gateway#api#openai-compatibleHow to Use Oh My OpenCode — Turn OpenCode into a Multi-Agent Team
Oh My OpenCode is an orchestration layer on top of OpenCode that gives you a team of async subagents. A practical guide to install, configure, run work, and stay safe on cost.
2026-07-06 · 16 min read #oh-my-opencode#opencode#ai-coding-agent#cli#developer-toolsLiteLLM in Practice — 100+ LLMs Behind One Interface
A hands-on quickstart for calling OpenAI, Anthropic, Gemini, Bedrock, OpenRouter, and local models through one OpenAI-shaped interface with LiteLLM — from install to streaming, Router, and the Proxy Server.
2026-07-06 · 9 min read #litellm#llm#ai-gateway#python#openai-compatibleMost Arguments Are About Ego, Not Ideas
When we argue, we believe we are contesting ideas, but we are often touching the other person’s sense of self. Through cognitive dissonance, identity-protective cognition, and the backfire effect, this piece explains why
2026-07-06 · 5 min read #communication#psychology#thinking#career