Andy Wheeler

I have a read a few of the books in those series, I think they are fairly well done -- both in terms of the topical material they cover and the quality of the summaries.

Combined cover pages LLMs for Mortals and Data Science for Crime Analysis with Python

I have a few different promo codes floating around the for the book. LLMDEVS gets you $25 off of the epub. TWOFOR1 gets you $30 off if you purchase multiple books.

crimede-coder.com/store

I have seen a few different groups use supervised models for this, so does not need to even be real fancy (could build one on CPU with historical changes).

Fellow criminologists -- is there anyone teaching gen AI courses in CJ programs? Let me know and will send them a copy of the book.

If no one is, feel free to get in touch. Happy to discuss doing a seminar/short course.

A couple of images from the old Nano Banana model I made for a presentation. I am not going to use these, but they are basically infographics out of the box.

This is a good reason to learn the LLM APIs -- this is not the case when you call the models directly instead of using the GUI interface.

It is particularly annoying to have LLMs write code to query their own APIs! I have them read documentation via web-search first if I notice an error, then resume.

Test driven dev helps with this as well, as the tests should be small, cheeky cases to illustrate the code (I understand with complicated code that still does not help, but could with this example at least). E.g. give a smaller sample with 5 words and have tests for each step.

Cover page, Large Language Models for Mortals

github.com/apwheele/LLM...

Buy the book to see more cool examples like this!

Pirate flag with KDE inside

Lets normalize using pirate flags instead of circular/hexagon logos for projects.

Created via a Gemini image MCP (to accompany Claude Code, since it does not have one). Link below

LLMs for Mortals, how to view the Epub YouTube video by CrimeDe-Coder

Not many people are familiar with epub, so I recorded a short video on how you can use free software to read the epub version of Large Language Models for Mortals

www.youtube.com/watch?v=OdJu...

Some notes on unreliability of LLM APIs Because my book, LLMs for Mortals, was created with Quarto, it runs the code when I compile the book. It uses cached versions when no code changes, but it is guaranteed to be working code for the p…

New blog post, some notes on unreliability of different LLM APIs I uncovered in my book. Have one example for OpenAI, Anthropic, Google, and AWS.

andrewpwheeler.com/2026/02/27/s...

If you are a machine and want a summary of what crime de-coder is working on, crimede-coder.com/llms.txt

New blog post, showing how to use log-probs to set the threshold for classification tasks (to get a specific recall or false positive rate)

Example toxic comment classification, even if True is under 1.2e-09, you should classify it as toxic to get 95% recall.

crimede-coder.com/blogposts/20...

Part of the rub with LLMs is that they are spiky (can work really well for one thing and be worthless for another), so knowing the tools well enough to test with your personal objectives are important.

Large Language Models for Mortals book released Large Language Models for Mortals: A Practical Guide for Analysts with Python

See my book for a walkthrough of using Google Antigravity, crimede-coder.com/blogposts/20..., as well as other coding tools.

Using Gemini.md to define how to set up conda for sub-agent

For the LLM tip of the day, to properly set up agents in the different LLM coding tools requires understanding python environments.

Claude code inherits from the terminal, Antigravity does not. But you can just tell Antigravity the correct entry point in `http://Gemini.md`.

I have a new book out -- Large Language Models for Mortals: A Practical Guide for Analysts with Python

This is an entry level book focused on using the foundation model APIs. Includes examples for OpenAI, Anthropic, Google, and AWS.

crimede-coder.com/blogposts/20...

I did a poll on LinkedIn that may be of interest to a few of my followers here. Would you prefer video/audio/written short form report. 1/3 preferred audio or video.

All college level knowledge workers.

Sponsor @simonw on GitHub Sponsors I'm building Datasette, LLM, sqlite-utils and a host of other related tools to make data analysis for small data (where small = "less than a TB") as productive as possible, and publishing a signifi...

I just sponsored @simonw. Go Sponsor your open source dependencies! You can join me at: github.com/sponsors/sim...

Part time product design positions to help with AI companies Recently on the Crime Analysis sub-reddit an individual posted about working with an AI product company developing a tool for detectives or investigators. The Mercor platform has many opportunities…

Recently someone shared a Mercor link on the Crime Analysis subreddit to help with AI product design. I collected a few different positions Mercor is recruiting for that may be of interest to my network -- writers, GIS, gov workers, and more.

andrewpwheeler.com/2026/01/15/p...

Planning an Experiment in Increasing Arrest Rates Planning an Experiment in Increasing Arrest Rates example in R

New blog post, R code showing how to plan how long to monitor an intervention intended to increase arrest rates

crimede-coder.com/blogposts/20...

Job info

Looking to hire part time help at Crime De-Coder. Get in touch or forward to interested students. crimede-coder.com/contact

Forecasts need to have error bars Richard Rosenfeld in the most recent Criminologist published a piece about forecasting national level crime rates. People complain about the FBI releasing crime stats a year late, academics are wor…

While I am open to more serious structural work, the current state of the field in those regards is pretty poor. Hence 30 years later, we still debate the great crime drop, and most crim work is "hey here is a new pet measure", andrewpwheeler.com/2023/11/19/f...

We don't know what causes wide spread crime trends

Blog post for the new year is a reprint of a Criminal Justician post I did, We don't know what causes wide spread crime trends.

crimede-coder.com/blogposts/20...

Year in Review 2025 and AI Predictions For a brief year in review, total views for the two different websites have decreased in the past year. For this blog, I am going to be a few thousand shy of 100,000 views. (2023 I had over 150k vi…

New blog post, year in review and putting together my predictions on whether AI will take all of our jobs.

Short answer no I do not think that will happen, and it may induce demand for more work in the end.

andrewpwheeler.com/2025/12/24/y...

We Hacked Flock Safety Cameras in under 30 Seconds. 🫥 YouTube video by Benn Jordan

Really important video about lack of physical security, and various lack of data security measures in Flock cameras.

www.youtube.com/watch?v=uB0g...

Coding Agents Risk Leaking Secrets Coding agents can leak secrets using exfiltration techniques. Here I tell you how to stop them or limit the harm.

New blog post on Crime De-Coder, coding agents risk leaking secrets.

I discuss two approaches to limit this risk. Whitelist IP were secrets can work from, or whitelist websites websearch tools can use.

crimede-coder.com/blogposts/20...

Advice for crime analyst to break into data science I recently received a question about a crime analyst looking to break into data science. Figured it would be a good topic for my advice in a blog post. I have written many resources over the years …

New blog post on advice for crime analysts looking to break into data science. Learn a programming language (python is most popular), and pick up some machine learning or LLM skills.

Those are the minimal skills needed for entry level data science jobs. andrewpwheeler.com/2025/11/21/a...

ArDrive Secure, permanent storage.

Another application for ARDrive and academics, pre-registrations

app.ardrive.io#/file/9badda...

Can see my grant with Arnold pre-reg data analysis plan, permanently available for everyone. I cannot delete this!