👋 Heya! I've been off Twitter for a while, but I'm planning on getting this back up and running for 2026.
I'm going to be posting about Python and software engineering topics to help you make better engineering decisions using Python.
I've been using Claude Code for software development for a while. If I had to share one thing about building with Claude it would be this...
It has a bias to over-engineer a simple prompt. If you include in your prompt "do not over engineer, keep it simple and scoped to what we discussed"
This prompt will save tokens, make the context window smaller so it's scoped to what you're trying to build, and make your code less complicated to review, build, and debug.
Over the course of my career, I've learned that dictionaries are one of the best data structures out there. They come built-in with a method that allows you to move away from bracket notation entirely, eliminating the need to avoid using try/except statements.
I sent it to over 1.7k subscribers on my email list, but you can see what this method is here: https://t.co/sFfnXT1Ruj
We should stop calling it “vibe coding” when it’s software engineers writing production software, paying close attention, thinking a lot, validating before committing, closing the loop… and yes they prompt an agent/work with it.
Very different from “let it rip” w 0 validation
It's painful when Python code runs slow. But there's 2 packages that help identify where code is not performing well: cProfile and line_profiler.
cProfile lets me see what functions are running slow and line_profiler shows me what lines aren't performing well.
I wrote about profiling about 6 months ago, but I might make a more up-to-date version: https://t.co/CPNkP252uz
Tomorrow, 1.7k+ newsletter subscribers will be the first to see what AI tools I'll be using for 2026 as a software developer. Hint: I don't use just Claude.
Subscribe to be the first to find out: https://t.co/m0gHK2TOhp
If this content interests you, I write a Python newsletter every week.
The first one of 2026 is coming out tomorrow, where I'm going to tell you how I'm using AI to code in 2026.
Subscribe for free: https://t.co/m0gHK2Um6X
👋 Heya! I've been off Twitter for a while, but I'm planning on getting this back up and running for 2026.
I'm going to be posting about Python and software engineering topics to help you make better engineering decisions using Python.
Over the last 2 years, I've been writing @Python_Snacks and this morning, I created my favorite thumbnail to date.
Instead of the usual "here's something cool you can do with Python", I wrote about some lessons I've learned over the years as a software engineer.
Officially launched the new @Python_Snacks website using the new @beehiiv website editor, and it is 🔥. Absolutely a well-needed upgrade.
Super exceited to continue building a resource hub for Python developers 😎
Officially launched the new @Python_Snacks website using the new @beehiiv website editor, and it is 🔥. Absolutely a well-needed upgrade.
Super exceited to continue building a resource hub for Python developers 😎
Exciting times in the world of newsletter data visualization.
- @beehiiv rolls out a new dashboard with deeper insights
- I'm continuing to build a dashboard for visualizing clicks (verified & by type) and generate advertiser reports.
My inner data nerd is happy.
I'll also add in: no hate towards Beehiiv or how they present analytics. The issue with CTR is if you have a bunch of corporate accounts clicking your average gets heavily inflated. Looking at verified is key, but Beehiiv doesn't present the verified data easily.
Weekend project: I've been using @AnthropicAI Claude to build a newsletter and advertiser platform for myself, and I am beyond impressed with the results.
The problem: the CTR presented on @beehiiv dashboard is heavily inflated. My solution 👇
No joke, this is a tool I'll forever use for my workflow. If you're a newsletter operator and you'd like to try it, @ me, DM me, etc.
It's in it's early development stages, but I have a vision.