Danny Penrose

Follow me for daily Tech, AI, ML & Productivity insights | CTO | AI & Automation | Digital Transformation | Chartered Manager (CMgr MCMI) | MSc AI

[Danny Penrose @dannypenrose]

Shipped search this week in a reading platform I've been building.

Full-text across title, author, notes, and highlights. Fast enough that it feels instant, even on a cold query.

The interesting decision was the index design.

I went with a denormalised search document per book rather than joining at query time.

More storage. Significantly better performance.

Worth it.

[Danny Penrose @dannypenrose]

Everyone's shipping AI code.

Most of it is not being reviewed properly.

Security holes, no error handling, queries that'll fall over under real load. Fast isn't the same as good.

I've spent a long time building engineering standards and agent skills that encode quality from the start, not as an afterthought.

Still doesn't replace an engineer who actually reads the output.

[Danny Penrose @dannypenrose]

7/ If you're building multi-tenant and haven't made this call deliberately yet, make it deliberately. Don't let it happen by accident.

[Danny Penrose @dannypenrose]

1/ Been working on the data layer for a personal project this week. Hit an interesting architecture decision around tenant isolation.

[Danny Penrose @dannypenrose]

6/ The thing about these decisions is they compound. Pick the wrong isolation model early and you're refactoring it when you can least afford to.

[Danny Penrose @dannypenrose]

Studying for an MSc while running a CTO role has taught me one thing above everything else.

I was confidently explaining a lot of things I only half understood.

Not wrong.

Just... incomplete.

There's a difference, and it matters when you're the person being relied on.

[Danny Penrose @dannypenrose]

4/ Cost me a day of rework to get there.

Worth it. Getting the data model right before the UI always is.

The lesson I keep having to relearn: abstract correctly first. The UI will thank you later.

[Danny Penrose @dannypenrose]

1/ Been reworking the forecasting engine in a personal finance project I'm building.

The tempting approach: aggregate all accounts into one running total and project forward.

It's wrong. Here's why.

[Danny Penrose @dannypenrose]

3/ So I split the model into separate projection tracks per account, each with its own rule set, then built the aggregate on top.

More complex to generate.

Much simpler to understand.

[Danny Penrose @dannypenrose]

Boards don't reject legacy modernisation because they don't care about technology.

They reject it because the CTO framed it as a technical problem rather than a commercial risk.

#TechnologyLeadership #CTO #DigitalTransformation

[Danny Penrose @dannypenrose]

@anglingdirect, I was just on hold for 30 minutes before being disconnected!! Frustrated is an understatement. Support tickets going unanswered, need help with several orders placed please.

[Danny Penrose @dannypenrose]

@corkulous You're not the only one, 3 of us in the household with pro, 2 had it a week and I use it more than anyone, and had a longer sub and still waiting. 🙄

[Python Visual Studio Code @pythonvscode]

📢 We’re excited to announce the general availability of Data Wrangler for VS @Code. Our new, code-centric, AI-powered data cleaning/prep tool is available to everyone today! 🔥

Learn more at https://t.co/A55MTKxCU9

[Danny Penrose @dannypenrose]

Ducked out after 30 days; pretty poor content, IMO and more of an affiliate marketing fest than anything else. Ok, for complete beginners.

[Danny Penrose @dannypenrose]

🧪 I'm excited to share that I'll be learning something new about AI for 100 days as part of the #100DaysOfAI challenge! Starting April 1st!

Join me via @100DaysAI!

https://t.co/dl5Br7umXH