Day 12: Success

Day 12

Today was fantastic! I had some time with my wife early in the morning sitting on our front porch and reading our Bibles. It was peaceful, calm, and focused on Him.

I spent the morning finishing my interview prep with some study items that were called out in my mock interview with ChatGPT. I read some PostgreSQL documentation and some React documentation, which, because I am a nerd, was very enjoyable. I’m constantly surprised at the sheer depth of PostgreSQL. For every feature it has, there are N sub-features and N^3 edge cases to consider depending on your data set and expected query shapes. One thing I’ve learned over the years is that your database experience largely depends on the problems you’ve had to solve and how much time you’ve had to solve them.

My interview went very well. It was two 1-hour blocks of pair programming with members of the engineering team. I got to be around my first “Principal Engineer” which was really cool. A core part of my ethos is to always look for someone who is better than me, consistently watch what they do over time, ask them questions, and then apply what I learn to my own workflow and technical knowledge. I’m excited about the opportunity to work alongside someone with that sort of experience, not to mention that the engineering culture at this prospective company seems highly collaborative and ego-less. That is an exciting combination!

I’m happy to report that I made it to the final interview in this process and will undergo reference checks in parallel. Stoked is an understatement. Blessed? Yea, very.

Ok, so I mentioned a YouTube video in my last post that changed my thinking and unlocked something in me. I can’t wait to apply a few of the learnings, but here are my bulleted thoughts:

• You must get yourself to a point where a dedicated agent sits between you and however many tasks you’re working on.
• Said agent must be able to give status reports of the work being done by other agents.
• You should be able to read status reports and understand which tasks need your explicit attention.
• The “definition of done” for agents must include a green CI pipeline, green tests, and green linting.
• You must determine a visual platform (TUI, IDE, whatever) that allows you to (a) interact with your status-manager agent and (b) interact with your individual tasks.

For me, the real “unlock” moment was the orchestration agent. I have been stuck at the point where I say things like “I can do maybe 3 things at a time”—I should be asking, “What else can I start?” An agent that can monitor other agents’ work is what will allow that to happen.

God Bless, folks.