Shep Bryan IV | Physicist and Innovator

Who killed Pip?

Mon, 11 Aug 2025 12:00:01 +0000

Chapter 1 — The Crime Scene

It was a quiet weekend in my project folder.

I had just created a fresh Python virtual environment — a private, self-contained folder that keeps all the packages a project needs in one place. I called it .venv.

Pip, the tool Python uses to install and manage packages, was alive and well. Everything worked… until I ran the most innocent of commands:

pip --version

But instead of a simple output, I was met with a chilling message:

ModuleNotFoundError: No module named...



Introducing CuentoGo
Fri, 25 Jul 2025 12:00:01 +0000
Vibe Coding a Spanish App
 A couple weeks back, my friend Julian Jacobs and I built CuentoGo, a Spanish learning app. Its a project I’ve been wanting to do for a while, but never had the time. That is until Julian introduced me to “Vibe Coding”. The idea in vibe coding is you just plan high level concepts and lean very heavily on AI to do all the grunt work. We decided to test it out and honestly, the results are pretty cool!
 
 The problem with vibe coding, or at least what people...


Building Train Hats
Fri, 18 Jul 2025 12:00:01 +0000
Building Train Hats
 In AI research, we’re trained to focus on improving the state-of-the-art. Many projects are designed to make incremental improvements with the hope that over long time spans, the small improvements add up. There’s value in that. But recently, I’ve discovered that the research projects that make the most immediate, tangible impact, aren’t the ones that incrementally improve models, but rather the ones that make existing models more usable. This was echoed frequently at the Mayo Clinic AI Summit I attended last week.
 Getting models into the real world is hard. And not in the ways we...


The Colors of Music
Fri, 07 Mar 2025 12:00:01 +0000
The Colors of Music
 Every note on a piano is defined by its frequency, for example middle A is 440 Hz. If you double the frequency, you get back the same note one octave higher. So 880 Hz is an A one octave up and 1760 is an A 2 octaves up and so on. If you keep doubling the frequency, eventually you will get to a frequency that is in the range of the frequencies of visible light. It follows that if there is a frequency in visible light that corresponds to the note A, then the color...


Turing Incomplete AI
Fri, 14 Feb 2025 12:00:01 +0000
Turing Incomplete AI
 I’ve been seeing a lot of discussion on current limitations of AI, but strangely I haven’t seen much talk about one of the biggest known limitations of current AI models – they are not Turing Complete. I wanted to talk a little bit about this, as it’s something AI enthusiasts and researchers should be aware of. Most of this discussion will be drawn from Ian Goodfellow’s Deep Learning book, which is truly a great resource for learning the basics of deep learning if you are looking for a good textbook.
 So what does it mean...


Shadowing Radiation Therapy
Fri, 17 Jan 2025 12:00:01 +0000
Shadowing Radiation Therapy
 This week I shadowed some radiation therapy treatments at the Mayo Clinic in order to see hands on what I’m helping build AI tools for. It was absolutely incredible seeing everything that I’ve learned about in research papers and textbooks play out in real life. More than that though, I just gained so much more appreciation for the power of human beings and what we have accomplished as a species.
 The treatments I shadowed were all performed on linac machines. The name “linac” is short for “linear accelerator” which is a method of speeding up electrons...


New Year Resolutions 2025
Fri, 10 Jan 2025 12:00:01 +0000
New Year, New Resolutions! I have high hopes for 2025, so to keep myself accountable, I want to make my resolutions public. If any of you catch me slacking, feel free to call me out. I encourage it!
 I’m especially excited about this new year because it coincides with the start of my new position as an AI Research Fellow at Mayo Clinic Phoenix. So my resolutions this year are largely career focussed. In particular I want to focus on building up my online presence and personal brand.
 Last year I started experimenting with LinkedIn and X to share my...


Physics, AI, and the Nobel Prize
Mon, 14 Oct 2024 12:00:01 +0000
A while back, I gave a presentation to my lab about the overlap between physics and recent advances in artificial intelligence. During the presentation, I made a speculative comment to add emphasis of the importance of the field: “Future Nobel Prizes in physics are going to be awarded to people who can advance the field of Artificial Intelligence.” When I said it, I had no idea how quickly that this prediction would come true.
 I want to be clear that I’m not trying to say I predicted the Nobel Prize or that I’m some futurist people should look up to....


Analyzing the NASDAQ with Gaussian Processes
Wed, 09 Oct 2024 12:00:02 +0000
A while back I worked on a fun project using some of the analysis tools from my PhD to analyze stock prices. Originally we just posted it as a GitHub repo, but I thought it would be fun to also post it here. So here it is! You can find the original repo here.
 Introduction
 The Gaussian process is a widely used tool in data science. The Gaussian process model is typically used for data smoothing, interpolation, and regression. However, there are other applications of Gaussian processes that make it a very powerful model.
 In this project,...


The Special Sauce at the Fall Festival
Wed, 09 Oct 2024 12:00:01 +0000
A fun story about my experience at a fall festival.
 The Special Sauce at the Fall Festival
 This past weekend I went to a pumpkin patch fall festival and they had a big slide that you go down in a tube. At the bottom there was a guy in a cowboy hat giving out tubes to ride the slide. I asked the guy what the trick is to go fast. He said to ask the guy in the red shirt at the top of the slide for the “special sauce”.
 When I got to the top of the slide...


Update September 2024
Tue, 03 Sep 2024 12:00:01 +0000
It’s been a while since I’ve written a life update, and a lot has changed since my last post. But, the dust has finally settled, and I am ready to report some updates.
 Moving
 This summer my family and I moved cross-country back to Phoenix. It more or less took a month to pack, a month to move, a month to unpack, and a month to settle in. We are living with my parents for a bit, while we sell our condo and look for a new place. But there are a lot of benefits to living at home,...


Mapping inference to physics - Part 3
Tue, 09 Jul 2024 12:00:01 +0000
This is the third installment of my project on mapping inference to physics. The previous parts looked at how we can map statistical inference of continuous variables to electrodyanmics (part 1) and how we can set up the general inference problem as a system of correlated binary random variables (part 2). In this third part we show that we can map the correlation between continuous variables and binary variables to a quantum mechanics.
 You can check out the latest version of the manuscript here: Information Physics. Keep in mind that this is still a work in progress,...


Mapping inference to physics - Part 2
Fri, 26 Apr 2024 12:00:01 +0000
A couple months ago I shared a post about a project we are working on in the lab. I got a lot of great feedback and I incorporated it into this new version of the paper.
 The idea of this project is to derive the physics of inference, in other words we want to try to derive equations for artificial general intelligence (AGI) models from first principles. The cool part about this project is that we are showing that model parameters of such a general inference algorithm can be mapped to physical quantities in such a way that when...


Comparing different models on biophysics data
Mon, 22 Apr 2024 12:00:01 +0000
Here I want to give a sneak peek into a project that my lab is working on. The goal of this project is to compare the results and performance of different deep learning models on a few biophysics datasets. We hope that this will be helpful, giving the community a better understanding of the strengths and weaknesses of different models on biophysics data, so we are structuring the code to be clean and easy to run. You can follow along with the project on GitHub, just please keep in mind that this is a work in progress.
 Background...


Update April 2024
Tue, 16 Apr 2024 12:00:01 +0000
Its funny how things can be going one way and then suddenly switch directions. Long story short, my family is moving back to Phoenix, and I’m going to start applying to jobs. I still plan to work on the startup, but I’ll be transitioning to a more part-time role.
 Moving
 Last year, my wife and I moved out to Florida so she could take a new job. Meanwhile, I have been working on a startup and doing some consulting work, taking advantage of the fact that her stable income would be able to support us while I worked on...


Update March 2024
Mon, 04 Mar 2024 12:00:01 +0000
Simultaneously energized by fatherhood and drained by lack of sleep, I spent last month catching up from my paternity leave. I’m now in a better spot than ever before and things are really looking up for Saguaro AI.
 Recently I started dividing my work into two main parts, plus a side project. The main parts are: 1) business development, and 2) the science itself. The side project is this information physics work I’ve been developing.
 Business Development
 What I am finding is that the most important part of a company is that it makes money. At the end of...


An ode to leap day
Thu, 29 Feb 2024 12:00:01 +0000
Leap day
 Today we celebrate Leap Day! It’s a special day partly because it is a rare day, but I think it is so much more. I think of Leap Day as a holiday celebrating the impact of science on society.
 In order for a society to have Leap Day, the society must have a team of dedicated astronomers studying the stars for an extended period of time who are smart enough to notice that the stars shift a bit each year. Then the society has to trust those astronomers enough to listen to them to add a day...


Information Physics Part One
Wed, 14 Feb 2024 12:00:01 +0000
This is a work in progress
 Typically when I have worked on projects in the past, I dont share them until they are complete. The advantage is that readers get a complete description of the project, but the problem with this strategy is that since writing papers is a long process, so I don’t get any feedback until the end of the project. Today I am going to try something different: I am going to share my work as I go. Keep in mind that since it is incomplete, there may be typos, errors, and logical pitfalls, but I...


Update February 2024
Tue, 06 Feb 2024 12:00:01 +0000
I am a dad now and it feels great! Its honestly like a power boost and I feel more motivated than ever.
 J. Shepard Bryan V was born on January 22. He was delivered via C-section and everything went really smoothly. He was 8 pounds 9 ounces and 22 inches long, which I am told is a big baby, though I have no frame of reference for baby sizes. So far he has been a very good and easy baby and we are really happy with him. My wife is also making a great recovery and is doing great.
 After...


Update January 2024
Mon, 08 Jan 2024 12:00:01 +0000
I started doing monthly reviews for friends and family as I continue down this startup venture. This is the second update I am posting on this blog, and it also coindides with the end of the year. So I will use this post to look back on 2023 and look forward to 2024.
 Year in Review
 I will remember 2023 as the year that I graduated with my PhD, officially started on this startup venture, moved to Florida, turned 30, and found out that I am going to be a father! This year was a big phase transition. I...


Building Backwards
Wed, 13 Dec 2023 12:01:00 +0000
Here we explore the business of biotech and how building backwards can help narrow in on a market niche.
 Background
 The first thing I learned when spinning out our university based research into a startup, Saguaro Solutions is that things that make great projects in academia are not automatically great projects in industry. It’s not that academic research is bad, it’s just that the goals are completely different.
 In Presse Lab, my PhD lab that specialized in building machine learning tools for biophysics applications, we were very diliberate about the projects we chose to work on because...


Chinese Character Generator
Thu, 07 Dec 2023 12:00:01 +0000
This article is a companion to the GitHub project. The project contains all the code necessary to run the model described in this article. The code is written in Python and is designed to be easy to read and understand. If you have any questions, please leave a comment below or message me directly.
 Introduction
 Generative AI for image generation is a booming topic right now. The field has rapidly advanced in an incredibly short amount of time, largely due to the discovery of clever ways to frame the image generation problem. In particular, diffusion based models have...


Update December 2023
Tue, 05 Dec 2023 12:00:01 +0000
It has been a little over 6 months since I graduated from my PhD program and started working on a startup. I have been providing regular monthly updates to family and friends via email, but now that I am actively using this GitHub blog I would like to shift to posting updates here.
 Since this is the first update I will give a quick recap of what we have been doing so far.
 How we got here
 I graduated with my PhD in physics from Arizona State University in May 2023, studying under my advisor Steve Presse. In Presse...


Dirichlet Clustering for Clinical Outcome Prediction
Wed, 15 Nov 2023 12:26:40 +0000
This article is a companion to the GitHub project of the same name. The project contains all the code necessary to run the model described in this article. The code is written in Python and is designed to be easy to read and understand. If you have any questions, please leave a comment below or message me directly.
 Introduction
 Clustering is an effective way of making predictions. It works by grouping data points together based on their features. Conversely, if we know the cluster a data point belongs to, we can make predictions about it. In medicine, this...