Being a Solo Dev In Current Environment

In this article, I’ll share my journey, a solo developer’s perspective. It’s been three months, and I’m still learning. The process has been surprisingly iterative, and I’ve discovered some valuable insights along the way. It’s a constant cycle of problem-solving, experimentation, and refinement. I’m focusing on building a solid foundation for Inkpilots. The initial stages were a bit chaotic, a whirlwind of ideas and half-formed concepts. Now, though, it’s starting to feel more structured, more manageable. I’m really concentrating on the core functionality to make sure my application does work quite wonderfully. Now system invalidates the previously sent requests, The initial chaos, however, is fading, replaced by a clearer sense of direction. The initial stages were a bit chaotic, a whirlwind of ideas and half-formed concepts. Now, though, it’s starting to feel more structured, more manageable.

Building New Features Blindfolded as Solo Dev

My latest feature was text completions on the core editor for Inkpilots. The new feature allows users to quickly generate text suggestions as they type, significantly boosting productivity and streamlining the editing process. I am testing the implementation right now to create this article. My words and LLM's words are quite mixed together, i know this may cause inconsistent voice. Especially i used the testing and a basic model for starting to text the functionality. Creating a feature is much easier with Github Copilot, i create a description prompt, then Copilot literally implements it for me, almost zero errors, It’s a fascinating interplay between human intent and AI generation – a truly remarkable experience. Since this model is quite stupid, results are not good enough, i should download a better model with higher training token count to test it. I can then see if model or the prompt has the problem. Later skaters. The model’s output is surprisingly consistent, a stark contrast to the unpredictable nature of human writing. The challenge, then, lies in aligning those two forces – the human’s creative vision and the AI’s algorithmic precision. It’s a delicate dance, and one that requires constant refinement. Latest sentence was written by Ollama local application. This feature was easier than i thought. Now, i increase the context length on Ollama application settings, now, it breaks for some reason. More i write, the content is not consistent, this is about model, it is too stupid to understand my intention, completions are related, but not enough for current llm standards.