● INSIGHTS
Lessons from real projects, written down.
Real project decisions and troubleshooting, across five topics we go deep on.
How should AI-DLC be applied to production maintenance?AX cases & methodology · August 24, 2026A practical guide to codebase readiness, task scope, version pinning, approval authority, verification evidence, and a 30-day pilot before applying AI-DLC to production maintenance.→
How much did Blacksmith Docker cache cut our development cycle? Evidence from 1,527 jobsCloud / AWS in practice · Updated August 12, 2026We examine how persistent Blacksmith builders, serialized cache writers, and parallel image and CI jobs changed production performance and development-cycle economics.→
Change the model or fix the prompt? An LLM selection evaluation on 200 production recordsEngineering deep dives · August 8, 2026We compared six configurations built from three models and three prompt stages on 200 production records without labeled ground truth. A model change added 14 percentage points of proxy agreement; a prompt revision added 37. The selected configuration costs $7.20 per month.→
Is private connectivity enough for LLM APIs? — Four boundaries from Bedrock to self-hostingRegulation & compliance · August 2, 2026A comparison of direct OpenAI and Anthropic APIs, Amazon Bedrock, Microsoft Foundry, Google Cloud, VPC deployments, and on-premises hosting across network path, processing location, retention, and legal basis.→
What should an AI gateway control? Enterprise requirements and implementation pathsCloud / AWS in practice · July 29, 2026A practical framework for putting security, key exposure, departmental cost, guardrails, and observability behind one enterprise AI gateway, with four implementation paths compared.→
How MCP 2026-07-28 changes AgentCore operationsCloud / AWS in practice · July 27, 2026How MCP's request-oriented model changes deployment and state ownership across AgentCore Runtime, Gateway, and targets, with a practical dual-version migration path.→
How far should AI content automation go? — Moving high-volume tagging and translation into productionAX cases & methodology · July 24, 2026How we validated manual content tagging and translation in a four-week PoC, then moved it into production at tens of thousands of items per month.→
An AWS Account With No Keys and No Docs — How $1,650/mo Became $400Cloud / AWS in practice · July 20, 2026We re-read a six-year-old service's abandoned AWS account through API inventory and measured metrics, then executed an Aurora major upgrade and rightsizing with zero-downtime tooling. Monthly cost dropped to a quarter without touching application code.→
A $455K AWS Cost Alert Arrived — The Response Made It WorseCloud / AWS in practice · July 19, 2026An AWS alert put one day of S3 data transfer at $455,000. The estimate was wrong, but we still had to investigate account compromise. The incident shows why billing communication is part of cloud security.→
Where should application modernization start? Map the transition before the rewriteApplication Modernization · July 18, 2026A practical guide to modernizing legacy applications without a big-bang rewrite: assess the current estate, choose retain, replace, replatform, refactor, or rearchitect, then design phased cutover and outcome metrics.→
Where does website blocking happen? Reading DNS, SNI, and CDN controlsEngineering deep dives · Updated July 18, 2026Map South Korean website controls across DNS, IP, SNI, and CDN layers, with July 2026 DoH, DoT, and ECH support and international cases.→
How do we evaluate RAG? A regression suite built from 30 questions and 8 follow-up setsEngineering deep dives · July 17, 2026We used 30 questions and 8 follow-up sets to isolate retrieval, grounded generation, and conversation context, then turn the results into RAG release gates.→
From AI PoC to production — how to design the transition gatesAX cases & methodology · July 16, 2026A six-gate method for moving an AI proof of concept into production. Decide with evidence across value, data, evaluation, risk, architecture, and operating readiness.→
Invisible on the Console, Only on the Invoice — AWS's Hidden CostsCloud / AWS in practice · Updated July 16, 2026Nobody broke in, yet money leaks every month. Missing VPC endpoints, cross-AZ data transfer, always-billed public IPv4, logs kept forever. AWS hidden costs with real Seoul-region rates and how to claw them back.→
Anatomy of an AWS Bill Shock — LLMjacking, and the Costs You Never SeeCloud / AWS in practice · Updated July 16, 2026One leaked access key can burn tens of thousands of dollars a day on Bedrock. How LLMjacking works, the AWS account-management essentials, Bedrock key management, and the guardrails that cap the blast radius.→
Why did one person fix in three days what eight people couldn't?Application Modernization · Updated July 16, 2026An eight-person SI project ended with core issues unsolved; alone, I fixed them in three days. Talent alone does not explain the gap, so I examine communication paths, diffused responsibility, decision structure, and the ownership gaps around re-architecture.→
Should you chase an agent-readiness score? From Level 1 to Agent-NativeEngineering deep dives · Updated July 16, 2026isitagentready.com scored our site 29, Level 1. Instead of filling the scorecard, we filtered by substance: Content Signals, markdown negotiation, a real MCP server, DNS-AID and DNSSEC — reaching 71, Level 5 (Agent-Native). And a hard look at what agent scores actually measure.→
From one chat line to a quote — a skill in the pipelineAX cases & methodology · Updated July 16, 2026A Claude skill that turns one chat line into a dev quote. Not a quote generator, but a human-in-the-loop skill inside the sales pipeline: the human keeps two gates (approve, send) while the skill handles the sheet, PDF, and email draft in between.→
Does encrypting SNI put traffic beyond control? Hiding ends at the CDNEngineering deep dives · Updated July 16, 2026ECH hides the destination domain from observers on the path, but it cannot change the policy decision of the endpoint that decrypts it: the CDN. How SNI and ECH work, which network layer controls what, and how the failure fingerprints differ.→
Did fully-managed RAG beat the pipeline we built?Engineering deep dives · Updated July 16, 2026We pit a hand-built pgvector pipeline against Amazon Bedrock's fully-managed Knowledge Base for a school-health advisory chatbot. On Korean table documents, what decided it?→
From a monolith to NKS: rebuilding the stack in four layersApplication Modernization · Updated July 16, 2026We moved a service locked into a single deployment and external CI onto NKS on Naver Cloud. How we split it into infrastructure, DB, API, and frontend — and made Git the single source of truth with Terraform, managed Cloud DB, and ArgoCD GitOps.→
● CONTENT PILLARS
Five things we go deep on.
Cloud / AWS in practice
Cost, configuration, troubleshooting — problem-solving records.
Application Modernization
From legacy assessment to phased cutover and operational handover.
AX cases & methodology
For decision-makers asking “how do we adopt AI.”
Regulation & compliance
For platform and healthcare regulatory practitioners.
Engineering deep dives
GPU, browser automation, on-device — high-difficulty keywords.
Turn the insight into a technical review of your challenge.
An engineer reviews your note first and, when useful, uses a 30-minute technical conversation to define priorities and execution scope.
● Already trusted by teams across finance · healthcare · media · public