Jane Hu

Hi, I’m Jane — a SWE exploring AI and agent systems 👋 I’m curious about how useful and reliable agents are actually built and shipped. I’ll share what I’m learning from papers, coffee chats, experiments, side projects, and the occasional random fun thing. I took a nonlinear path into tech, studied across 4 fields in 7 years, and didn’t take any CS classes in undergrad — so I know what it feels like to learn from scratch, and I care about making technical ideas easier to understand. If you’re new to CS, AI, agents, or coding, welcome. If you’re already deep in the field, I’d love to learn from you too. Let’s learn together :)

@feifei_qiu Thanks :)

@janehu07 Can't wait to see what you build

Hi, I’m Jane — a SWE exploring AI and agent systems 👋 I’m curious about how useful and reliable agents are actually built and shipped. I’ll share what I’m learning from papers, coffee chats, experiments, side projects, and the occasional random fun thing. I took a nonlinear path into tech, studied across 4 fields in 7 years, and didn’t take any CS classes in undergrad — so I know what it feels like to learn from scratch, and I care about making technical ideas easier to understand. If you’re new to CS, AI, agents, or coding, welcome. If you’re already deep in the field, I’d love to learn from you too. Let’s learn together :)

@slowAndNeat yea let's go, feel there’s still a gap between traditional SWE and building with AI

@janehu07 Fellow Swe! Great to see ... On my way of understanding these stuff too !

@EricCLFung Lol

@janehu07 Just start paying for some tokens, and you are off the races

@EmmaHLeonhart nice to meet you too :)

@janehu07 Nice to meet you! I too have taken a nonlinear path into AI. Previously studied economics and philosophy.

@johnfurr538500 let's connect!

@janehu07 Hi Jane, Robotics and AI is my main focus. Happy to connect with people entering the space.

@billionsoffuck I double majored in math and psychology in undergrad, and studied ds & cs in grad school :)

@janehu07 What’s your background Jane?

Seems like nonlinear paths into tech resonate with a lot of people. If you’re interested in hearing more, on a similar transition path, or just want to connect, feel free to like or comment :)

@techn0eacc I double majored in math and psychology in undergrad, and studied ds & cs in grad school :)

@janehu07 What were your other fields of study?

Your path is honestly inspiring! I barely coded in undergrad and only started learning seriously in grad school, so I relate to finding your way in from an unusual angle. Tbh school and industry can feel like totally different worlds. A degree helps in some ways, but it’s clearly not the only path. Let’s go :)

What do you mean when you say “nonlinear paths into tech”? Due to many various factors, I’ve always deeply wanted to get a degree and join academia but just never got even close. Because of this I’d had to try so many different ways to brute force my way into the dev world; pushing as much open source projects out nonstop is what eventually got me noticed enough to bypass needing a degree and I now work to get completely unrestricted localy run agents into the hands of the common persons. I’d say this counts a pretty “nonlinear” path.

@JamesLaneAI 💪

@janehu07 I come from an IT help desk back ground but now I'm leaning into AI development and evaluation

@tech_optimist let's go!

@janehu07 It does indeed! Took a very nonlinear journey into the field myself. Having a different perspective than the typical CS researcher has its pros, and it's great to see more diverse minds enter the field. Let's go 💪🏽🚀

@feynman_ri82211 Thank you! Still very much learning in the trenches too, but I’ll share whatever I figure out along the way.

@janehu07 Probably more like you will be teaching and sharing to us, looking forward to. Welcome to the trenches. Anyway congratulations on graduating

@chrisreedbates great insight!

@janehu07 I think infra / systems engineering / agent orchestration are the coolest thing to look at right now… it’s like building cars - you can build a ford astro or a Ferrari. Both are cars.

@erd0xbc team codex!

@janehu07 Codex >>>>>>

Very useful Claude Code coaching video from qiuzhi2046: youtube.com/watch?v=r-CGDA… Wish I had watched this a month earlier 🥹

@joshzjs Likewise! Really enjoyed the conversation. Looking forward to following what you’re building at e2a.

Great chatting with you Jane!

I had a coffee chat with @joshzjs , founder of e2a, about his experience of building agents @Google and why AI agents may need their own software tools. If you don’t want to read the full article, here are my main takeaways: 1. AI may become a new kind of operating system, with tools and infra built around it. 2. When building agents inside a large company, interacting with legacy systems is hard because much of the domain knowledge lives in senior engineers’ heads. 3. Infra skills transfer well into agent building because LLMs are still APIs inside larger systems. 5. Generic agent apps may get absorbed by foundation model companies, but agent infra still has room. 6. e2a is betting on software tools for agents, not humans. The goal is to give agents their own identities so they can operate more like human employees, with their own email, phone numbers, and other work tools. If you’re curious about what Josh is building, check out e2a here: e2a.dev He’s also happy to connect with people working on agent infrastructure or developer tools. @joshzjs

@pete80155006 thanks, this means a lot 🥹

@janehu07 The biggest advantage of AI is that people who don‘t need to use it are so professional. I am also a computer layman, but in just two months, I have been proficient in using tools such as Claude code to greatly improve productivity. So you must succeed, Jane..

@sido2038 thanks :)

@janehu07 welcome Jane I’m sido.

@AIGC0to1 thanks :)

@janehu07 加油加油！

@an_engineer_log haha fair, I’m not a native speaker so I use AI to polish my wording 🤫 but the thoughts are mine

@janehu07 those em dashes though 🤣

@stevencheng thanks :)

@janehu07 welcome to the space. shipping reliable agents is mostly edge case wrangling. been doing similar vibe coding for robot vision, keen to follow your experiments.