Pradeep Gowda btbytes

Karpathy

A few random notes from claude coding quite a bit last few weeks.

Coding workflow. Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write... in words. It hurts the ego a bit but the power to operate over software in large "code actions" is just too net useful, especially once you adapt to it, configure it, learn to use it, and wrap your head around what it can and cannot do. This is easily the biggest change to my basic coding workflow in ~2 decades of programming and it happened over the course of a few weeks. I'd expect something similar to be happening to well into double digit percent of engineers out there, while the awareness of it in the general population feels well into low single digit percent.

IDEs/agent swarms/fallability. Both the "

Conversation Messages with Claude Code

All user messages from this Claude Code session:

on individual article pages, put the date below the title of the page, and remove the date from the footer. keep the permalink, tags in the footer.
merge bin/post.py and bin/link.py into bin/new.py and allow a non optional first param to make a post or link or note or quote.. use the title and url as next positiona params
for note use "note" as the slug, unless there are more than one note in the same day, in which case increment the numbers like note2, note3 etc

A python script that when run will look at the files committed to git staging and uses a openai api key from the zsh environment and write a commit summary, and use that to commit the changes.

Uses uv to make this a standalone script.

Written using Claude.

I use this to automatically generate git commits to my blog source code.

My Observation on our Employee Survey results

To: All AT&T Managers

Thank you to the over 99,000 employees who responded to this year's Employee Survey. Your feedback helps us understand what we need to focus on—and how we need to communicate — as we strive toward our goal of ranking in the top third of Fortune 100 companies in employee engagement. It is incredibly encouraging that 73% of our employees took the time to respond to the survey, with 79% of those respondents feeling committed and engaged with their work at AT&T. While this is reassuring, especially considering the amount of change we've navigated as a company recently, it wasn't a surprise to me that we fell short of our engagement goal. Nor was it a surprise that scores declined since our 2023 survey. We are midstream on a multi-year journey to build the company we want, not simply optimize the one we have — and I believe your feedback accurately reflects where we are.

In the survey you asked for transparency in communication from leadersh

Asked Claude to:

write a one page pamphlet about moby dick using postscript.

Convert the PS to PDF using:

ps2pdf mobydick.ps

| | Hacker News new | past | comments | ask | show | jobs | submit | login
---|---|---
| 1. | | We spent $20 to achieve RCE and accidentally became the admins of .mobi (watchtowr.com)
---|---|---
| 681 points by notmine1337 5 hours ago | hide | 164 comments
| |
2. | | AppleWatchAmmeter (github.com/jp3141)
| 117 points by rcarmo 3 hours ago | hide | 14 comments
| |
3. | [](vote?id=41511928&h

Unwitting React

20-Hour React Study Plan for those who hate it

1. React Fundamentals (3 hours)

Components and JSX
Props and state
Lifecycle methods (not required if you use functional components - hook)
Hooks (useState, useEffect)

	#!/usr/bin/env python3
	"""
	Create an Azure AI Search vector index and ingest a directory of PDFs.

	Requires:
	pip install azure-search-documents pypdf openai python-dotenv

	Env vars:
	AZURE_SEARCH_ENDPOINT
	AZURE_SEARCH_API_KEY

	# Sample text
	text = "the cat sat on the mat the dog sat on the floor"

	# Tokenize the text
	words = text.split()

	# Build bigram model
	bigrams = defaultdict(list)
	for i in range(len(words) - 1):
	bigrams[words[i]].append(words[i + 1])

	<!DOCTYPE html>
	<html lang="en">
	<head>
	<meta charset="UTF-8">
	<meta name="viewport" content="width=device-width, initial-scale=1.0">
	<title>blunderfest</title>
	<script src="https://cdn.tailwindcss.com"></script>
	</head>
	<body class="bg-gray-100 font-sans">
	<div class="max-w-3xl mx-auto p-8">