Cloud Nirvana Cleveland is in the books
I had been stressing out about this one for a while.
Earlier this month, I gave a keynote at Public Sector Network's Government Innovation Ohio event and sat on a panel afterward. It was an in-my-element talk, given the government innovation theme. It didn't go perfectly. I planned to speak from the podium, expecting a screen for my notes, and found out once I got up there that the notes screen was in the middle of the stage. I ended up talking to the slides without my notes crutch.
We also did that whole “what kind of questions do you want to field” thing before the conference, and I got some curveballs. But it was still good!
Cloud Nirvana came a couple of weeks later; this sort of tech conference prizes the tech and how its implemented. It's a different kind of room:
- It's aimed at practitioners and goes deeper.
- It asks speakers to cover both wins and challenges.
Cloud Nirvana wants to hear about the work in the field, so I gave them my mental model of where data teams have to operate.
This is how I think about building and maintaining a gold standard data function at a highly regulated organization. You start with dependable data engineering. If you keep that layer rock solid, it holds you out of the moat of bad practices.
I covered much more, but the point I most wanted to land is that backsliding is always possible. There is no permanent removal from the moat. A few failed pipelines erode trust in the system, and strength anywhere else won't stop the slide.
This is something practitioners are better placed to say. Behind the promises of brand new MBAs at tech consulting outfits, pop-up niche vendors, and so on, the reality is that this work is hard and continuous. Let’s say your engineer is taking an extended vacation… are your practices strong enough to hold for that period of time?
If you're having core issues with data quality, your ELT processes, and getting basic reporting to the right people, sales pitches for semantic layers, AI, and decision models are super-premature.
