Blake Werlinger: Origins (lmao)

This is my first writing on my portfolio page! Very exciting! The writing in these posts might not be as professional as LinkedIn, I’m just seeing how it feels to write about myself. I think where I want to start is to go back through my history in programming. I did something similar with my previous portfolio website, but it was a lot of AI assisted writing back then and I’ve since decided that AI generated writing really has zero place in a website that’s meant to express who I am.

All that is to say, this post is a fun opportunity to reflect on where I started, make up for the old version of this history story in my previous portfolio, and most importantly I want to shake off a lot of the imposter syndrome that comes with the AI coding era.

I think the best place to start is with Scratch (https://scratch.mit.edu) A platform that I first found at probably 10 - 11 years old. I came in wanting to make games, and I guess you could argue I did. With their little art editor and block based programming I made MS Paint looking Five nights at Freddy’s clones and cookie clicker games that had a playtime of maybe 10 minutes. I vaguely remember making dozens of games and it taught me the foundations of programming as a whole. Loops, conditions, variables, the whole shabang. I’m not entirely sure if I would have gotten into programming if I hadn’t stumbled across Scratch.

That interest eventually grew into a big motivation for game development, which at 14 led me directly into Unity Game Engine. Unity gave me a whole new world to work in, and after a couple months of following tutorials I felt like I was genuinely able to write novel C# scripts. I didn’t yet know that you could create classes that didn’t inherit from Monobehavior (or what inheriting or classes even were) but I did know that you could set variables and control them inside of Start() or Update() methods across different GameObjects. I got obsessed with Unity for the better part of 2-3 years. I would probably make 3 projects a week and finish zero because my ideas were either “I’m going to remake GTA but low poly” or “I don’t really have any idea how to impliment this basic thing, let me fiddle with it”.

I remember spending hours once making a water visual by layering a bunch of opaque blue planes and being able to change the distance between them and being so proud at how good it looked. I hadn’t heard of shaders and I just assumed all game VFX were just layering semi transparent planes or something.

By 17 I had gotten a handle on the real concept of OOP and started working on video game mods. I created a couple mods for BTD6 which was fun because it’s a game that while made in Unity is compiled in IL2CPP. I did skip a lot of the hard part of IL2CPP by relying on a community mod helping library (though I did eventually contribute to that mod helper so I will give myself some credit) The main mod I remember was a tower that I added which was based on the Banana Farmer. To spare the details for people who don’t know what BTD6 is or why I’m mentioning bananas, I’ll just say that I’m very happy with the attention that the mod got. A popular youtuber ISAB even made a video on my mod that got ~1.4 million views which was a very nice surprise. ISAB Video

I jumped around a lot of modding stuff, Rimworld was a game I got very into and did a few mods for, Rivals of Aether had a few mods that got popular. Around that time there was a short period where I started experimenting with creating hacked clients in video games but even at 18 where I thought I was being edgy it just felt like it was too much and I wasn’t really getting anything out of it. And the games I thought I was cool for hacking were random steam games and some of them weren’t even multiplayer, so I think I just thought it was cool to do DLL injection regardless of what I was doing it for.

In my senior year of high school I finally did some real expansion of my programming ability. I learned Python because I wanted to make my own little version of Jarvis. This was late 2021 early 2022 so pre-ChatGPT and LLM’s weren’t good enough to be interesting. I still got around to doing some ML stuff for it though. I took a NLP pytorch setup that someone else made for intent recognition in text so I could turn statements into commands, and got to train it on more samples to expand it’s ability to recognize intent. It was basic but it was what I needed to convince myself how easy and fun it was to learn other languages besides C#.

Then November 2022 hit and ChatGPT launches and it’s still weird to think about how that changed everything. I immediately wanted to create an AI assistant in my terminal to chat with that would use GPT 3.5 and vividly remember being upset at how long it was taking for them to add GPT 3.5 to the API. So I was on it pretty much day 1, not to hype myself up too much since I’m pretty sure saying that devs were interested in AI after ChatGPT came out is an understatement.

I would know the scale of the dev reaction because almost immediately after GPT 4 comes out in March of 2023, I get to hop into a discord of a little project called AutoGPT. I would have the opportunity of joining an internal group of contributors, and then watch as a few days in the repo goes viral and hit 100k+ stars on github. The project was super simple compared to what you’d expect from a big AI project in 2026, but in 2023 the idea of an AI model in a loop of writing out JSON for a command, and then running that command was mind bending. Even though it would get stuck running the same command over and over again because it was designed for conversation flow with a human only, when it worked I think that we all saw the vision of what was to come. I wasn’t a huge code contributor on the project, but I did get to be on the inside and talk to the founder/original maintainer of the project, Toran Bruce Richards (Significant-Gravitas), and loved being part of standups at 18 for an Open source project. I was doing R&D of ways to handle tokens on such small context windows (you guys have to remember that GPT 4 was originally like 8k tokens). I also was recording demo videos, one of which actually got posted on the official README.md of the project which looking back is still very suprising. I am super grateful to Toran and the rest of the talented people who I got to meet on that team who let some 18 year old kid demo their project for 100k+ people watching. AutoGPT has evolved a lot since I worked on that team and I don’t really have any contribution to it’s modern state, but I do keep up and have came in with a couple PRs even recently.

Since then I’ve shipped a couple small things. But I’ve mostly been using my understanding, alongside the ever increasing power of LLMs to make dozens of personal projects, some of which I release, most of which stay as little tools for my day to day.

I won’t go into too many of my modern projects in this post since I think each of them deserves their own blog post. I appreciate whoever decided to read through everything I wrote here. I know it might seem kind of random for me to outline every switch in language, but I just felt like this is a good spot for a developer portfolio blog to start out. Stretching the muscles of talking about myself instead of asking Claude to take a list of things and fill in the blanks. As I mentioned before, this blog will have zero AI influence, which is why you might see some typos, some actual signs of life maybe. I’m writing this at 3 AM so me being alive right now is debatable. Thank you for reading!