Skip to content
View miles-howell's full-sized avatar
🥽
Trillin
🥽
Trillin

Block or report miles-howell

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
miles-howell/README.md

Hi, I'm Miles

I enjoy working on low-level problems, mainly in AI and cybersecurity. In highschool, while OpenAI was still focused on Gym, I wrote a single-neuron network and reworked Karpathy's famous min-char-rnn to show off to classmates.

Also in highschool, I co-lead the programming club and consulted for the state's first Cybersecurity high school cirriculum.

In college, I spent 2 years doing coursework in Computer Science and Mathematics. During my time there, I started and lead the ML/AI Special Interest Group (SIG). This was an extracurricular activity that was optional to anyone at the school. At the time, there were no undergraduate ML/AI courses available and only a handful of graduate-level courses available. In the sessions, I broke down complex topics in ML using plain English and no high-level libraries.

The goal was not to build or design a product or tool, but to understand the technology at its core. High-level libraries outsource your understanding! Obviously things have changed in AI development, and writing a production codebase of any kind with no libraries would be insanity. But this time was invaluable to me in that I fully understood the concepts behind machine learning.

Personally and currently I'm interested in the frontier of AI research and experimenting with new implemenations. I've gotten incredibly bored with the big AI labs companies and their current pace of research. If not pace, then style and focus of their work. I miss the days when anything was possible with AI!! Now, we only call it a "breakthrough" if it means my AI waifu uses less tokens to tell me she loves me.

Really though, I'm just exhausted of LLMs and the news cycles. There are much MUCH more interesting (and world-altering) ways to use this technology. We just started to scratch the surface when ChatGPT and its counterparts appeared and forever changed the landscape of ML research, skewing everyone's perspective towards a narrow use case of a technology that's existed for nearly century!

Email me at miles@plannedpixel.com to chat or discuss my work.

Pinned Loading

  1. Bridge-Crossing-RL-Sim Bridge-Crossing-RL-Sim Public

    A Django web app for visualizing an RL model learning to build a bridge and make it home.

    Python 1

  2. shrimple shrimple Public

    Make posture training "shrimple"! A tiny shrimp sits in your Windows task tray, and gets mad when you don't check your posture.

    Rust 1

  3. Logarithmic-Hamming-Numbers Logarithmic-Hamming-Numbers Public

    A geometric & logarithmic solution to solving for the nth Hamming Number, allowing for faster calculation of very large Hamming Numbers compared to Dijkstra’s three-pointer method

    Python

  4. Vixxing Vixxing Public

    Quick and convenient way for IT teams to verify their user's identity over the phone using a familiar TOTP registration.

    HTML

  5. fruit-classifier fruit-classifier Public

    An educational example of binary classification using a basic neural network built without ML libraries.

    Python 1

  6. Window-Works Window-Works Public

    A small Django app for tracking an office floorplan. Intended as a quick reference for IT teams and others who need to know an employee/computer's location.

    JavaScript 6