Articles

Notes from shipping AI products and backends

28 practical articles on LLM apps, RAG, backend engineering, DevOps and the craft of working as an engineer — written from production experience, not from the docs.

Vibe Coding

AI-paired engineering. Claude Code, Cursor, agents, prompts.

Editors & Agents The tools that turn an editor into a collaborator.

Vibe Coding · 4 min Vibe coding with Claude Code How to delegate, when to drive, and what changes about software engineering when an agent is paired with you. Vibe Coding · 3 min Cursor: the editor as conversation An editor where chat is a first-class citizen and inline edits replace the autocomplete stream. Vibe Coding · 5 min Agents and MCP: the long-running coding session When 'one chat, ten edits' becomes 'one prompt, a whole feature.' What MCP actually does, and why it's the protocol I expected without realizing I expected it. Vibe Coding · 4 min Aider, Continue, and the CLI-first workflow Editor-integrated AI is great. CLI-integrated AI is sometimes better. When and why I reach for Aider instead of Cursor.

Workflow & Prompts How to actually drive these tools day-to-day.

Vibe Coding · 4 min Prompt engineering for engineers Stop calling it 'engineering' until you treat it like one. Versioning, eval-driven iteration, and the prompt patterns that survive contact with production. Vibe Coding · 5 min A practical scorecard for AI coding tools How to actually compare Cursor, Claude Code, Aider, Continue, and Copilot. The criteria that matter, the ones that don't, and what changes between teams.

AI / ML

LLMs, RAG, embeddings, fine-tuning, evaluation.

Foundations The mental model engineers need before shipping LLM features.

AI / ML · 7 min LLMs, explained for engineers who use them daily What's actually happening inside that API call. Tokens, attention, context windows, sampling — the parts that change how you write code with them.

Retrieval & Embeddings RAG, vector search, hybrid retrieval — without hand-waving.

AI / ML · 5 min RAG patterns that actually work Naive top-k similarity is where most RAG systems start and where most of them fail. Here's what to do instead. AI / ML · 5 min Embeddings + vector DBs without the magic What an embedding actually is, how vector search works under the hood, and how to pick between pgvector, Qdrant, and Pinecone.

Tuning & Eval Fine-tuning, prompt tuning, and how to know if any of it works.

AI / ML · 5 min When fine-tuning beats prompt engineering Most fine-tuning attempts are premature. Here's how to tell when you actually need it, and what cheaper alternatives to try first. AI / ML · 5 min Evaluating LLM apps: past the demo The gap between 'looks good in the demo' and 'works at p99 in production' is where most LLM apps die. Here's how to measure your way across it.

Operating LLMs Cost, latency, fallback, observability for LLM apps.

AI / ML · 4 min LLM cost & latency: math before scaling Back-of-the-envelope math, the levers that matter (caching, batching, model choice), and the failure mode that bankrupts new LLM features.

Backend

Postgres, Redis, APIs, distributed systems, ops.

Databases Postgres + Redis, the workhorses behind everything.

Backend · 4 min Postgres: the 12 settings that matter After tuning a hundred Postgres instances, these are the dozen knobs I actually touch. Everything else is mostly noise. Backend · 5 min Redis: cache, queue, lock, mistake — pick one Redis is fantastic at four things and dangerous when you mix them. A pragmatic field guide for production use.

APIs & Auth Designing and securing your service interface.

Backend · 5 min API design without the cargo cult REST vs RPC vs GraphQL is the wrong question. Here's a checklist that keeps you out of the worst API design mistakes. Backend · 5 min Auth: JWTs, sessions, OAuth, what to actually use A decision tree for picking the right auth approach. JWTs are not the answer to every question, sessions are not obsolete, and most teams over-engineer this.

Distributed Systems Queues, consensus, idempotency, the things that bite at scale.

Backend · 3 min Distributed systems for backend devs Idempotency, retries, and the eight fallacies that bite you when one service becomes ten. Backend · 5 min SQS, RabbitMQ, Kafka — when to pick which Three different jobs, three different tools. The decision tree, the pitfalls, and what to do when 'just use Postgres' is the right answer.

Operations Observability, caching, the "how does it run in prod" layer.

Backend · 4 min Observability: traces beat logs Logs find what you knew to look for. Traces find what you didn't. The pragmatic case for OpenTelemetry as the foundation of production debugging. Backend · 4 min Caching: where to put it, how it bites you back Caching is the easiest way to make a slow system fast and the easiest way to make a correct system wrong. A field guide to the layers.

Personal Branding

Career, money, freelance, remote, GitHub, networking.

Online Presence LinkedIn, GitHub, blogs, networks — your public surface.

Personal Branding · 5 min Building a LinkedIn presence that doesn't feel like LinkedIn Show your work. Skip the inspiration porn. The audience is smaller and more valuable than you think. Personal Branding · 5 min Your GitHub is a portfolio Most engineers' GitHub profiles are accidental. The ones that earn jobs and credibility are intentional. A short guide to making yours useful. Personal Branding · 6 min Technical writing as a career multiplier The most under-priced skill in software. How to start writing publicly, what to write about, and how the compounding works. Personal Branding · 5 min Networking for engineers who hate networking If 'building your network' makes you cringe, you're doing it wrong. The version that actually works for introverted engineers.

Career & Money Salary, freelance, remote — the engineer's income engine.

Personal Branding · 7 min Making money as a software engineer (without selling out) A taxonomy of income streams — salary, contracting, products, royalties — with rough numbers from someone who's tried most of them. Personal Branding · 7 min Freelancing as a backend engineer — the real playbook How to find clients, what to charge, what to refuse, and how to avoid the loneliness trap. Personal Branding · 7 min Remote work, done well What actually changes when you stop sharing a room with your team — and the habits that keep you sharp instead of slowly drifting. Personal Branding · 7 min Salary negotiation for engineers (with scripts) The compensation gap between 'engineer who negotiated well' and 'engineer who didn't' is six figures over a career. The exact moves to make.

Prefer to learn this with me? I train engineers and teams in full-stack, DevOps and AI, one-to-one or in groups.

Book an appointment WhatsApp