AI Learning and Development

rewrite shorter but still in my voice: I am trying to solve that problem myself. I have been studying for quite some time and still I do not feel prepared. The way I have earned money from artificial intelligence is mostly through data annotation, which I outline here: https://danielkliewer.com/blog/2024-11-27-data-annotation-guide So what I have experience in is data annotation. It is not exactly easy to get into and does not necessarily pay well. I have found some gigs which have paid well but they are the exception rather than the rule. Yet more and more companies continue to pop up as demand for LLMs increased. But to answer what I taught myself it went kind of like this: First I was interested in data science. So I started taking intro classes for data science from edx.org from Harvard and intro to computer science classes on edx.org from MIT. In the process I started learning fundamentals from ocw.mit.edu I already knew calculus and basic prob/stat so I learned more advanced prob/stat and linear algebra as well as data structures and algorithms. Although I started with R, I soon moved mostly to Python and taught myself more and more software engineering standards through developing things with Python. So I learned FastAPI, Flask and Django. At this point I realized that my front end understanding lagged backend so I took the Odin Project to refresh on HTML/CSS and JavaScript. I had a small background in javascript but nothing serious. After that I started learning TypeScript for React, Vite, and Next.js Around this time the first LLMs came out. After that my learning both accelerated and decelerated depending on how you look at it. Now my React projects actually worked. Around the same time I restarted my blog, danielkliewer.com , I started integrating LLMs into my projects. You can see the progression on my website of what I taught myself from there. What I know how to do now is to integrate LLMs into full stack applications. I almost created a product for someone which would have been difficult but possibly doable and launched a freelance career, but I chickened out, because I was not confident enough in my abilities. I just did not feel confident enough. So instead I am trying to build more projects, to develop better templates, create frameworks for larger projects, learn more software engineering standards, there are so many of them. I am getting closer and closer to my initial goal, that is using my knowledge I taught myself to support myself. Currently I still work two jobs, one in data annotation and the other is a manual labor position which keeps me healthy but very tired. I would like to transition to only earning from developing AI solutions for people. That is my goal. I still have more to learn. My learning rate is slower than before I got the manual labor job, but I am much more financially secure now than when I was hopping from contract to contract doing annotation. I try to stay active in subreddits. That seems to help with learning. I try to create things and receive feedback. It is just like being an artist, receiving criticism and building better things. It takes thick skin, which I don't always have, but it is part of the learning process. So the answer is two-fold. One I got immediate employment without real qualifications through working in an adjacent industry to artificial intelligence, that is data annotation. That work helped me tremendously and is what helped me get housing. Two I have been teaching myself software engineering directly related to developing full stack applications integrated with LLMs and machine learning in hopes of finding full time employment, which is still in progress. Hope this helps in some way.

I’ve been tackling this problem myself. Despite studying for a while, I still don’t feel fully prepared. Most of my AI-related income has come from data annotation, which I detailed here: [Data Annotation Guide](https://danielkliewer.com/blog/2024-11-27-data-annotation-guide). It’s not easy to break into and rarely pays well, though I’ve found a few exceptions. As LLM demand grows, more companies keep emerging. My self-taught journey started with data science, taking intro courses from Harvard (edX) and MIT (OCW). Since I already knew calculus and basic probability/stats, I advanced into linear algebra, data structures, and algorithms. I began with R but switched to Python, picking up FastAPI, Flask, and Django. Realizing my frontend lagged, I took The Odin Project, then learned TypeScript, React, Vite, and Next.js. When LLMs emerged, my learning shifted. My React projects started working, and I integrated LLMs into full-stack applications. Restarting my blog, [danielkliewer.com](https://danielkliewer.com), helped track my progress. At one point, I nearly took on a complex freelance project but backed out due to self-doubt. Now, I’m focused on building projects, refining templates, and mastering software engineering standards. My goal is to fully support myself with AI development. I still work two jobs—data annotation and manual labor—which keeps me financially stable but exhausted. Transitioning to full-time AI work remains my priority. Learning has slowed since taking the labor job, but I’m more secure than when I relied solely on contract work. I stay active on Reddit, create projects, and seek feedback—it’s like being an artist, growing through critique. In short, I got my start through data annotation, which helped me get housing, and I’ve been self-teaching full-stack AI development to land full-time work. Still a work in progress, but I’m getting closer. Hope this helps.