Skip to content
Home>Blog

Inside Engineering

Architecture decisions, delivery lessons, and the trade-offs that shape complex systems.
An abstract image representing a big computer, showing streams of gold lines descending from above.

Optimising data pipelines: sometimes you just have to stop and think
Video

What is RAG in AI? Reducing LLM hallunications with retrieval-augmented generation
Video

How vector embeddings help AI understand language context
Video

How do large language models (LLMs) really work?
The backs of a woman and man at their desk in an office

Why developers (and everyone else) should care about analytics
Video

What are all these Data X words? Warehouses, Lakes & Lakehouses

Are LLMs destroying the planet?
Video

What actually is data mesh?
Video

What is data lineage?

AI natives: beyond the hype of modern dev tooling 
Video

How to make the most of metadata

Why delivery managers should think like designers
A picture of a gate post with lock on it and the gate ajar.

Stop calling it “Prompt Engineering”
A wooden signpost with multiple badges on it, each of which point in different directions to different locations.

An opinionated guide to writing expressive tests

The CrowdStrike IT Outage: 3 key takeaways to prepare for the next big platform disaster
a picture of a surprised child

Let’s talk about invisible tech debt! Vol. 2: Surprise dependencies (and key steps to tackle them)

A plea to Engineering Managers: Let’s talk about invisible tech debt!

When open-source libraries close up: replacing core dependencies

Data should provide your questions, not just your answers

Diving into the deep blue data lake, maths in 3072 dimensions, and the perfect recipe for Data Spaghetti

What’s up with all these “Data-X” words?
This image is a flame graph representing the performance profile of a software application. It is primarily colored in gradients of red, orange, and yellow, with various shades indicating different levels of resource consumption. The vertical axis represents the stack depth, where each block represents a function in the call stack at a moment in time; the wider the block, the more time the program spent in that function. Text labels overlay the blocks, indicating function names, with some labels mentioning 'waitress', 'pyramid', 'bootstrap', and 'our code'. The image is complex, with many overlapping layers of blocks and text, suggesting a detailed profiling session designed to diagnose performance issues or bottlenecks within the software.

A love letter to flame graphs: or why the perfect visualisation makes the problem obvious

How to turn your sophisticated spreadsheet models into enterprise-class tools

Lessons learned building big datasets with GCP Dataflow

Three ways simple Kanban boards make your projects easier to run 

Container Orchestration: Lightweight Container Runtimes 

Injecting the pizzazz (back) into your retrospectives

IaC roundup: Microsoft Azure Bicep

Service teams: Five ways to improve your accessibility testing

Mobile app development: When is React Native the right choice for your business?

Alpha projects: How do you make yours a success?

Three stages to MDM-enabled data maturity

Four definitions of ‘good design’ that we love

How to scale an image processing pipeline to hundreds of workers using Kubernetes and KEDA

Discovery projects: What do they involve and how do you make them a success?

DNSControl (Infrastructure as Code series)

What really happens when a Kubernetes container runs out of memory?

Four workarounds to use when Scrum or Kanban agile methods don’t work

Designing an operating model to unlock the potential of a headless CMS build

How to choose the right Container Orchestrator
An image of somebody typing on an illuminated keyboard.

How to spot the signs you’ve outgrown spreadsheets

Building resilience into your technical team

How can strong data governance improve your AI project?

Five ways to get the most from your analytics project

Arithmetic using the Scala Type System

Implementing a custom Netlify CMS widget