I’m non-technical in the traditional sense: I went to school for finance and took a few introductory computer science courses, but nothing else. When I started my first coding journey back in college, I learned a lot about the lowest levels of data structures and algorithms, but the education seemed geared towards landing its students FAANG jobs instead of building novel projects. As I’ve spent the past two years in venture capital, I’ve had the privilege of meeting many talented engineers and scientists, all capable of coding and building. While my building capacity was limited to a financial model at best, these people were building real things that others would actually use. And so, I took it upon myself to learn to code and play around with open source software in an effort to improve my skills and most importantly figure out how it all works.

Therefore, I’ve been spending a lot more time on GitHub. Sure, I’ve been thinking on new projects and seeing what I could build (such as the website you’re reading this on), but I’ve been drawn into an innocuous button in the navigation menu. The Trending page features the most popular publicly visible projects from the GitHub community. Many of the repositories are filled with useful information or tooling for developers, but many of them research from top labs and institutions. This is where I’ve fallen down the rabbit hole.

GitHub offers a portal into research to see what the smartest people are working on and what the future of technology may look like. Through GitHub (and by extension arXiv), I’m able to peer into the world of academia. Much of this research is concentrated in LLMs given it’s the prevailing zeitgeist. Many of the top open source AI labs were founded by former academics, so it makes sense that building in public and forming communities around projects has become the default.

GitHub is one vector through which a non-technical person like myself can be introduced to new research and technology developments. However, it’s up to me as the reader to decide which teams are worth following, so when they commercialize or disband, I’ll be on the frontline of new companies and products. I believe that following academic developments should be in every venture capitalist’s toolkits, as it’s not about where we are, but where we’re going. A few of my favorite recent repositories are linked below.

AgentGPT

Eureka: Human-Level Reward Design via Coding Large Language Models

OpenAgents: An Open Platform for Language Agents in the Wild

GitHub