DevSecOpsSecurity Automation Developers Do Not Route Around
A hardcoded credential sat in the security report for three weeks, beside 190 low-severity notices about test fixtures. The tooling was fine. The signal was not.
Read MoreThe Blog
Page 4 of 4 — notes on engineering, AI and the cloud.
AllDatabases 4API Development 3AI Agents 2AI Coding Tools 2Artificial Intelligence 2DevOps 2Freelancing 2Performance 2AI Agents vs Chatbots 1AI Automation 1AI Engineering / Machine Learning 1AI Infrastructure 1AI Observability 1AI Security 1Authentication 1Backend Development 1Blockchain 1Career 1Cloud Computing 1Cloud Security 1Context Engineering 1Cybersecurity 1Data Engineering 1Data Science 1DevSecOps 1Docker 1Edge Computing 1Emerging Technologies 1Engineering Metrics 1Feature Flags 1FinOps 1Generative AI 1Git 1Interviews 1IoT 1Kubernetes 1Large Language Models 1MCP 1MVP 1Machine Learning 1Message Queues 1Microservices 1Node.js 1Observability 1Platform Engineering 1Prompt Engineering 1Quantum Computing 1RAG 1React 1Real-Time 1Robotics 1Rust 1SRE 1SaaS Architecture 1Small Language Models 1Software Development 1System Design 1Testing 1TypeScript 1Vibe Coding 1Web Development 1Web Fundamentals 1WebAssembly 1Zero Trust 1
DevSecOps
Artificial IntelligenceChatGPT vs Gemini vs Claude: How I Actually Choose in Production
Most teams argue about which model is best. After shipping generative AI for real clients, I've found the model matters far less than the workload around it.
Read More
Edge ComputingEdge Computing and 5G: When Moving Compute Closer Actually Pays
A client wanted edge computing for a slow page. Network latency was 180ms of a 2.3 second load. The rest was a missing index and 1.8MB of JavaScript.
Read More
IoTIoT Architecture That Survives 5,000 Devices
The fifty-sensor pilot worked perfectly. At five thousand devices every assumption in it turned out to be wrong — not slightly wrong, wrong in kind.
Read More
BlockchainBlockchain After the Hype: Where a Ledger Actually Beats a Database
I have been asked to build on a blockchain four times. Three would have worked better as a Postgres table with an audit log. The fourth was a genuinely good fit.
Read More
DockerDocker From First Principles: Containers, Compose and Production
A release slipped two days because a server had Node 16 and a laptop had Node 18. Nobody was careless — environments drift, and that is what containers fixed.
Read More
KubernetesKubernetes Without the Fog: The Mental Model I Wish I Had Started With
Exit code 137, no useful error, two hours reading application code. The container was being killed for exceeding a memory limit somebody copied from a template.
Read More
MicroservicesMicroservices vs Monolith: The Question Is Not Which, It Is When
Six microservices, four people, forty requests per second. Every feature touched three of them. They had split up because monoliths supposedly do not scale.
Read More
API DevelopmentREST, GraphQL and the API Design Decisions You Cannot Undo
We shipped a list endpoint without pagination because the list had thirty items. Eighteen months later it had 62,000, and four clients depended on the shape.
Read More
Data EngineeringETL, ELT and the Pipeline Failures Nobody Warns You About
A pipeline ran green for eleven months while dropping 4% of records. Nobody found it through monitoring. An analyst found it because a number felt low.
Read More
AI Coding ToolsCursor, Copilot or Claude Code: Which AI Coding Tool for Which Job
I used whichever tool was already open for six months and got inconsistent results. The variation was not in the models — it was in the shape of the task.
Read More
RAGRAG in Practice: Vector Databases, Chunking and Why Retrieval Is Your Problem
My first RAG system answered confidently and completely wrongly in front of a client. The model was fine. The right paragraph was never in the context.
Read More
Prompt EngineeringPrompt Engineering for People Who Ship: Patterns That Survive Production
The best prompt improvement I made last year was deleting three paragraphs. Accuracy went from the low eighties to the mid nineties.
Read More
Large Language ModelsWhat an LLM Actually Is: Tokens, Context Windows and Why Your Model Forgets
A developer spent a day debugging a summariser that kept ignoring the end of long documents. There was no error. The context window was simply full.
Read More
Machine LearningSupervised, Unsupervised, and the ML Projects That Actually Ship
A team spent four months on a churn model that was never used. Not because the model was bad — because nobody asked when the predictions would be needed.
Read More