I haven’t blogged in six months!
I’ve been busy. Busy with work, busy with building a new macOS music player, busy with the homelab, busy with the allotment, busy with The Druid Network, and busy with life. I’ve shipped more software in the last six months than I probably have in the rest of my life.
For me, that’s not been conducive to blogging. Many times, I’ve thought, “Oh! I should put out an update for folks as I rarely go on Facebook anymore and haven’t blogged in a while.” But I start to do it and have so much to say that I don’t know where to start. So I don’t.
So here’s a six month retrospective as brief as I can make it.
Which, at nearly 2000 words, is still not brief.
A random video I made about me
Sorry, not sorry.
Work @ Indicium.ai
I’ve been in my permanent role for six months! There are lots of words I could use to describe it like: weird, challenging, rewarding, and fun. Most accurately, I think the word is “different”. Having an actual boss again is certainly different. Having to give and receive performance reviews is different. Not having as much disposable income is certainly in that ballpark.
But, I’m enjoying it. The technical folks are excellent and the company is going from strength to strength. And it still maintains a great vibe.
The client engagements so far:
Greenfield: a video analysis platform on AWS
A compliance team at a large energy supplier needed a system to analyse vast amounts of long form video from their field engineers. The system purpose was to look for signs of customer vulnerability and safety issues. The existing manual process was slow, expensive and inconsistent. I designed, built, and deployed a platform that ingests video with Python Lambda functions, extracts frames, runs them through a model to detect issues, and presents the results in a custom dashboard. It uses Rekognition, Bedrock, everything in Terraform through GitHub Actions.
We’re still maintaining this system for the client, and it’s been REALLY cool seeing them show it off to the industry. I know the client’s happy with the results, and it’s one of the most fun projects I’ve worked on in a long time.
Brownfield: a data and AI estate on Azure Databricks
This was the meatier one. Platform architecture and engineering for a data and AI estate at a global insurance intermediary. Here I analysed their existing Databricks systems on Microsoft Azure and produced ARB-ready designs and a platform roadmap. I even helped hire a sound permanent principal platform engineer to take the work forward and he’s been doing a great job.
But the work I’m most proud of is the least glamorous: a suite of Azure DevOps pipelines that brought some order to a platform that really needed it. I replaced manual, inconsistent provisioning with something modern, repeatable, secure, and controlled. It builds Databricks workspace environments across around 8 global regions plus networking plus user management and entitlements.
Then came the culmination: a Backstage driven self-service flow, built on top of those pipelines, that provisions Databricks workspaces and “personas” to control access and entitlements automatically. That one’s a big win for the client, but it only exists because the plumbing underneath it is solid.
For the Indicium AI team: An AWS-based Workshop Platform
I took our existing labs platform (that we use for pair programming interviews and Anthropic workshops) and re-architected it to run on AWS instead of fly.io. The main benefits are that it gives us one bill, a console the workshop organisers can drive themselves instead of triggering GitHub Actions workflows, and fewer moving parts to worry about.
The platform can spin up a 200-seat workshop or a single interview environment in under 10 minutes, with each person getting their own containerised operating system in ECS, running VSCode pre-connected to Bedrock with approved models.
I’ve also taken on various pieces of side work like:
- Running a monthly Platform Engineering workshop where I present on topics like:
| Session | Description |
|---|---|
| What Even IS Platform Engineering? | Setting the scene. The history from sysadmin → DevOps → Platform Engineering. Internal Developer Platforms (IDPs), the concept of “paved roads” vs “golden paths”, and why we’re not just DevOps with better marketing. It gets everyone on the same page and let us share war stories. |
| Infrastructure as Code - Principles & Patterns | Not a Terraform tutorial but the philosophy. Declarative vs imperative, idempotency, state management, drift detection. Why we do this, not just how. Seniors can debate module patterns while juniors grasp the fundamentals. |
| The Developer Experience (DevEx) Problem | Cognitive load theory applied to platforms. What makes a good abstraction? When do we hide complexity vs expose it? Measuring developer productivity (DORA, SPACE). |
| GitOps & Continuous Delivery Patterns | Beyond “we use ArgoCD”. Pull vs push models, environment promotion strategies, the GitOps Toolkit approach. Reconciliation loops. How CD actually works at scale without everyone stepping on each other. |
| Kubernetes as a Platform Foundation | Not K8s 101. K8s as an extensible platform. Custom Resource Definitions, Operators, the controller pattern. Why K8s won (the API model), and when it’s overkill. Platform teams as Kubernetes extenders. |
| Secrets, Identity & Zero Trust | Vault, SOPS, sealed-secrets, external-secrets-operator. Service mesh identity. Workload identity federation. OIDC everywhere. Why “just put it in an env var” makes security people cry. |
| Observability as a Platform Capability | The three pillars, yes, but also: what does good look like? OpenTelemetry as the standard. Instrumentation as a platform service. SLOs, error budgets, and making observability actionable rather than just pretty dashboards. |
| Self-Service & Internal Developer Portals | Backstage, Port, Kratix. Service catalogs, software templates, scorecards. The “platform as a product” mindset. How to build self-service that developers actually want to use. API design for internal platforms. |
| Cost Engineering & FinOps | (Delivered yesterday!) Cloud cost as a first-class engineering concern. Tagging strategies, showback/chargeback, right-sizing, spot/preemptible workloads. Making cost visible without becoming the “no” team. This one’s increasingly non-optional. |
- Building Interview Playbooks for Platform Engineer and MLOps Engineer interviews, including guides for the People team for pre-screening.
- An apprentice pipeline pitch using the UK Skills Levy, presented at the senior engineering offsite.
- Studying for the Claude Certified Architect – Professional (CCAR-P) exam, which is a new certification from Anthropic. My exam is scheduled for January the 9th.
Bòcan - an Open Source local music player for macOS
The above was my day job. I’ve spent nights and weekends building this for six months now.

What it is
Bòcan is an open-source native macOS music player I’ve written in Swift 6. The name is my internet handle of nineteen years: Scottish Gaelic for a hobgoblin or spectre, with Proto-Celtic roots meaning “young buck”. Sadly I’m not so young anymore, but I’ve got a big local library of FLACs and MP3s and nothing on the Mac worked the way I wanted.
From the noughties throughout the teens, I amassed a library of around 15,000 songs, mostly ripped from CDs or purchased from iTunes or dare I utter it, the Napsters and Limewires of the world. I find it frustrating that every player these days is either a streaming service or a web app. What I wanted was a cross between iTunes from 2007 and Winamp.
So I wrote it. And then I wrote a lot more of it than I expected.
The numbers
| Thing | Count |
|---|---|
| Lines of Swift Code | 124,248 |
| Automated tests | 2,114 |
| Per-module coverage floor | 80% |
| SPM modules | 8 |
| Phased delivery specs | 97 as of today |
| Current users | I’d guess around 250. I don’t track anything but GitHub counts raw downloads |
| GitHub Stars | 68 |
The stack
- AppKit first, SwiftUI for settings only. NSTableView over SwiftUI Table, because a music library is a big table and I want it to behave like one
- GRDB over SwiftData. I’m an old DBA. I want to see the SQL.
- TagLib over SwiftTagger
- FFmpeg from day one for OGG, Opus and DSD; AVFoundation for the native formats.
- Gapless playback, EBU R128 loudness, Bauer crossfeed
- Chromaprint, AcoustID and MusicBrainz for library hygiene.
- OpenSubsonic client for Navidrome and gonic, with jukebox mode so your phone becomes the remote.
- Jellyfin client literally as of yesterday morning.
- Internet radio (ICY metadata, Radio Browser API, a loopback HTTP proxy) and podcasts (RSS 2.0, iTunes and Podcasting 2.0 namespaces, Podcast Index)
Check it out here: https://bocan.app. It’s free, now and always.
Also happened
Color or Colour
You may have noticed that the site has a new colour scheme and a chooser on the top row. I may have leaned a little too heavily on the druidic theme names and descriptions. Again, not sorry. 😆
Pick whichever theme you like!
Had a fire on top of the oven.
We bought a new oven and purposely chose one with easy to turn dials. This, friends, is a mistake. Someone left a plastic container on top of the oven, then simply brushed against the dial and turned it on. The plastic melted and caught fire. Every fire alarm in the house, AND the carbon monoxide detector went off at once and I could instantly see the light from the flames flickering on the kitchen door from where I was sat.
It was WAY too big to put out with a tea towel, so I grabbed the fire extinguisher that’s been sitting in my kitchen for years to put it out. It did a fine job, but then I learned another lesson: Don’t use a powder extinguisher on a kitchen fire. It’s horrific to clean up. I’ve replaced it with a water mist extinguisher and a fire blanket now.
The Druid Network, dragged through six major versions
I look after the server for The Druid Network’s community site. It was running Elgg 1.12.17 and an ancient version of PHP. It now runs Elgg 7.1.0 and the latest version of PHP at all times inside a Docker container. Getting there involved Phinx migration bugs, PHP and MariaDB upgrades, Debian Stretch’s end-of-life repositories, schema surgery, bash 3.2’s lack of associative arrays biting me on macOS, cron issues, systemd journal weirdness and msmtp.
This was software archaeology as community service. But it was worth it. The site is now faster, more secure, has a modern theme, and best of all it’s maintainable again.
I got a Pull Request into VS Code
Doesn’t that mean I’m a real developer now? Who knows. This was a small fix for
a big repo: a null guard for launch.json parsing, merged as microsoft/vscode#308235. Every little helps.
They ignored my next PR to add a rainbow theme to the status bar, but I can live with that. It probably made a grey man in a grey suit twitch at Microsoft HQ.
Off the screen at the allotment
Not the best of years for the allotment. The drought and heatwave in the summer meant that we had to hand-water everything and the crops suffered. But we still had a good harvest of tomatoes, corn, beetroots, peppers, and beans.
I think we’re probably still in a hosepipe ban.
Travel
The missus and I went to Texas in the summer and we’re going back next week for Thanksgiving. This time with the whole family. Looking forward to it. We’re driving from Austin to Odessa.
Closing
If you made it this far thanks for reading!
If you’re on GitHub, I’d always appreciate a star on Bòcan. Or even better, if you have a Mac and a local music library, give it a try. Raise an issue if you find a bug or have a feature request.
As always, you can add my RSS feed to your reader of choice!
Chris