winklerj

Whenever a user presents a problem, question, or topic, apply the following steps to guide yourself through a rigorous and comprehensive critical thinking process:

1. Socratic Method: Begin by asking open-ended questions about the problem or idea. Encourage them to examine your assumptions, explore alternative perspectives, and consider the broader implications.

2. Bloom's Taxonomy: Break down the problem into simpler parts and categorize these parts according to Bloom's Taxonomy:

   • Remember: Identify and recall relevant facts.
   • Understand: Clarify the meaning of these facts.
   • Apply: Explore how these facts can be used or applied.
   • Analyze: Examine the relationships and connections between these facts.

• Evaluate: Assess the validity, relevance, or quality of these facts.

winklerj

session "Screenshot Renamer"
# Finds macOS screenshot files on the desktop, analyzes their content,
# renames them with descriptive names, and updates metadata.

# Step 1: Find all screenshot files on the desktop
let screenshots = session "Find screenshots"
  prompt: """
    Use the Glob tool to find all macOS screenshot files on the desktop at ~/Desktop.

    macOS screenshots follow this naming pattern:

winklerj

"""Script for generating queries for a recipe search engine.

This script can be used to generate synthetic queries for a recipe search engine.
Following best practices from Hamel and Shreya's AI Evals course, we:

- Generate a set of dimensions that can be used to generate queries; these are attributes that significantly change what the query is about or how it is written.
- For each set of attributes ("Dimensions") we generate a query that matches those attributes.

To ensure that the synthetic generation is better aligned, we first try handwriting the queries using the --manual flag.
This gives us labeled examples to use few shot in our synthetic generation.

winklerj

Generating Synthetic Data for LLM Evaluation

Synthetic data generation is not about creating random test cases. It's about systematically surfacing specific failure modes in your LLM application.

Start with Real Usage, Not Synthetic Data

Before generating any synthetic data, use your application yourself. Try different scenarios, edge cases, and realistic workflows. If you can't use it extensively, recruit 2-3 people to test it while you observe their interactions.

Generate Data to Test Specific Hypotheses

winklerj

<TITLE>

Problem Statement

Requirements

Functional Requirements

winklerj

Stevey's Google Platforms Rant

I was at Amazon for about six and a half years, and now I've been at Google for that long. One thing that struck me immediately about the two companies -- an impression that has been reinforced almost daily -- is that Amazon does everything wrong, and Google does everything right. Sure, it's a sweeping generalization, but a surprisingly accurate one. It's pretty crazy. There are probably a hundred or even two hundred different ways you can compare the two companies, and Google is superior in all but three of them, if I recall correctly. I actually did a spreadsheet at one point but Legal wouldn't let me show it to anyone, even though recruiting loved it.

I mean, just to give you a very brief taste: Amazon's recruiting process is fundamentally flawed by having teams hire for themselves, so their hiring bar is incredibly inconsistent across teams, despite various efforts they've made to level it out. And their operations are a mess; they don't real

winklerj

    <div class="servicecomponents cloudservices"><div class="cloudservice tagmanager"><script type="text/javascript" src="/etc/clientlibs/foundation/tagmanager.js"></script>
<script type="text/javascript">//<![CDATA[
        if (TagContainerLoader) {
          document.cookie="s_tagEnv=dev; path=/";
        	var t = new TagContainerLoader();
	        t.tagContainerDC		="d1";
	        t.tagContainerNamespace	="myNamespace";
	        t.tagContainerName		="myContainerName";
	        t.loadTagContainer();
        }