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
Deno 2.9 — The deno desktop Experiment, Lockfile-Preserving Migration, and a 24-Hour Supply-Chain Default
Deno 2.9, released on June 25, 2026, comes down to three stories: deno desktop, an experimental feature for building native desktop apps with the web stack (OS webview by default, an optional bundled CEF); a migration pa
2026-07-17 · 14 min read #javascript#deno#javascript-runtime#node-compatibility#supply-chainHow AI Agents Fail in Production — 14 Failure Modes, and Why Retrying Is Not Safe
Putting agents in production hurts in three places. First, failures come from system design, not the model — UC Berkeley's MAST study classified 1642 execution traces into 14 failure modes, and 44.2% of them were system-
2026-07-17 · 20 min read #ai#agents#observability#reliability#mcpAI Coding Agents: What to Use for What — Selection Criteria Verified Only Against the Four Vendors' Official Docs
When you are choosing among Claude Code · Cursor · GitHub Copilot · OpenAI Codex, the question asked most often is "which one is cheapest." Yet the prices these four companies publish cannot answer it — because all four
2026-07-17 · 24 min read #ai#ai-coding-agent#claude-code#cursor#github-copilotIs AI Code Review Actually Worth It — What Measured Evidence Says About Accuracy and False Positives
Marketing copy for AI code review tools is full of numbers like "80% of PRs need no human comment," yet almost none of it publishes precision and false-positive rates alongside. Pull the public measurements together and
2026-07-17 · 25 min read #ai#code-review#static-analysis#evaluation#software-engineeringCassandra 6.0-alpha1 and Accord Transactions — Where Does the 5-Year "General-Purpose Transaction" Promise Stand Now
Accord, the general-purpose transaction protocol for Cassandra proposed as CEP-15 in 2021, finally shipped in a runnable release, 6.0-alpha1, in March-April 2026. The protocol promises to handle multi-partition strict-se
2026-07-17 · 13 min read #database#cassandra#distributed-systems#transactionssystemd v261 — Staged Rollout Comes to PID 1, Cloud IMDS Absorption, and a Finished dlopen Migration
systemd v261, released 2026-06-19, ships ConditionFraction= (turns a unit on for only a set fraction of the fleet), ConditionMachineTag= (splits machines into rings via machine tags), and a new IMDS subsystem that pulls
2026-07-17 · 16 min read #linux#systemd#devops#cloud#sysadminWho Maintains OPA Now — Styra's Team Joins Apple, and 11 Months of Verifiable Record
On August 20, 2025, an announcement said OPA's founders and a number of Styra team members were joining Apple, and a rumor spread through the community that 'Apple acquired OPA.' Checking primary sources tells a differen
2026-07-17 · 12 min read #security#opa#authorization#open-sourceMaking Video from a Single Image — Kling·Veo·Sora vs Wan·HunyuanVideo, What to Pick and When
When you are choosing a model to turn a single image plus a prompt into video, what actually decides it is not the polish of the demo reel but three things: price per second, input-image constraints, and licensing. This
2026-07-17 · 24 min read #ai#video-generation#image-to-video#open-weights#licensingContent-Encoding: zstd — Safari 26.3 Fills In the Third Engine, and What's Next Is RFC 9842 Dictionary Compression
In February 2026, Safari 26.3 added support for Content-Encoding: zstd, so all three browser engines now receive zstd responses, following Chrome 123 (2024-03) and Firefox 126 (2024-05). Because it's negotiation-based, a
2026-07-17 · 12 min read #performance#web#network#browserHow AI Agent Memory Is Actually Built — Four Designs, and What the Benchmarks Really Prove
"Agent memory" is not a single technique but a blur of at least four distinct designs — a file scratchpad, summarization/compaction, vector retrieval, and knowledge graphs. This post pins down what each actually does fro
2026-07-17 · 21 min read #ai#ai-agent#agent-memory#llm#benchmarkAfter Airflow 2 EOL — The Real Work List for Going 2 to 3, and Where 3.x Stands at 3.3
Apache Airflow 2 passed EOL on April 22, 2026, and it no longer receives security patches. But the road from 2 to 3 is not a pip upgrade — it is an architecture change. Direct metadata-DB access from workers disappears,
2026-07-17 · 15 min read #data-engineering#airflow#workflow-engine#migrationZJIT's Lightweight Frames — How Ruby 4.1 Erases Method-Frame Pushes, and What It Costs
Ruby 4.0 shipped ZJIT on December 25, 2025 — but shipped it turned off: it's compiled into the binary, yet the default JIT is still YJIT. The official release notes put it this way: "ZJIT is faster than the interpreter,
2026-07-16 · 20 min read #ruby#zjit#yjit#jit#compilerNative Async in WASI 0.3 — Why wasi:io Disappeared, and What's Still Not Done
On June 11, 2026, the WASI Subgroup ratified WASI 0.3.0. The core change is that async moved down into the Component Model's canonical ABI, and as a result the wasi:io package disappeared entirely. This matters not for p
2026-07-16 · 19 min read #webassembly#wasm#wasi#component-model#systemsSimulated Customers Never Walk Away — Where LLM User Simulators Inflate Agent Scores
In τ-bench-style conversational agent benchmarks, the 'user' role is played by another LLM. But this simulator isn't the thing being measured — it's the measuring instrument, and instruments need calibration. Three verif
2026-07-16 · 21 min read #ai#llm#evaluation#agents#simulationTriton Gluon: The Language That Hand-Writes the Layouts the Compiler Used to Hide
Gluon is a low-level GPU language built on the same compiler stack as Triton — it hands kernel authors the layouts, shared memory, and warp specialization that Triton normally keeps hidden. Its reason for existing is cle
2026-07-16 · 19 min read #gpu#triton#kernel#compiler#performanceThe Rust Standard Library Verification Campaign Found Zero Memory-Safety Bugs
The results paper for the verify-rust-std campaign, led by AWS and the Rust Foundation, appeared in the 2026 NASA Formal Methods Symposium. The work — what the authors call 'the largest verification campaign ever reporte
2026-07-16 · 25 min read #formal-methods#rust#verification#memory-safety#testingThe SQLite WAL-Reset Bug: A Data Corruption Race That Hid for 15 Years
On March 3, 2026, SQLite developer Dan found and fixed a data race that could, rarely, corrupt a database in WAL mode. The bug lived in every version from 3.7.0 (2010-07-21), when WAL was introduced, through 3.51.2 (2026
2026-07-16 · 18 min read #sqlite#wal#database-internals#concurrency#data-raceZig 0.16.0 — The Release That Turned I/O into an Interface, and the Bill It Came With
Zig 0.16.0, released on April 14, 2026, changed the standard library so that every I/O operation takes an Io instance as an argument. It takes the old Zig habit of passing an Allocator as an argument and applies the same
2026-07-16 · 23 min read #zig#systems-programming#async-io#compilerTypeScript 7 Is GA — the 8-12x Is Real, but Some People Cannot Move Today
On July 8, 2026, TypeScript 7.0 went GA. The Go-rewritten native port is no longer a separate binary — it is now simply tsc, and npm install -D typescript pulls it in. Microsoft's reported full-build speedup is 8 to 12x,
2026-07-16 · 18 min read #typescript#compiler#tooling#build-systems#compile-timesOctober 11, 2026: The Root KSK Changes — What's Actually in the Root Zone Right Now
The root KSK — DNSSEC's trust anchor — moves from KSK-2017 to KSK-2024 on October 11, 2026. For most resolvers this is a non-event thanks to RFC 5011, but for a validating resolver that failed to pick up the new trust an
2026-07-16 · 25 min read #network#dns#dnssec#security#protocol