← Back to context

Comment by fallinditch

5 months ago

My guess is that, yes, the software development job market is being massively disrupted, but there are things you can do to come out on top:

* Learn more of the entire stack, especially the backend, and devops.

* Embrace the increased productivity on offer to ship more products, solo projects, etc

* Be highly selective as far as possible in how you spend your productive time: being uber-effective can mean thinking and planning in longer timescales.

* Set up an awesome personal knowledge management system and agentic assistants

13 comments

fallinditch

Reply
shinycode  5 months ago

We have thousand of old systems to maintain. Not sure everything could be rewritten or maintained with only LLM. If an LLM builds a whole system on its own and is able to maintain and fix it then it’s not just us software developper who will suffer, it means nothing to sale or market, people will just ask an LLM to do something. No sure this is possible. ChatGPT gave me a list of commands for my ec2 instance and one of them when executed made me loose access to ssh. It didn’t warn me. So « blindly » following an LLM lead on a cascade of instructions on a massive scale and on a long period could also lead to massive bugs or corruption of datas. Who did not ask an LLM for some code, that contained mistakes and we had to point the mistakes to it. I doubt system will stay robust with full autonomy without any human supervision. But it’s a great tool to iterate and throw away code after testing ideas

whynotminot  5 months ago

> Learn more of the entire stack, especially the backend, and devops.

I actually wonder about this. Is it better to gain some relatively mediocre experience at lots of things? AI seems to be pretty good at lots of things.

Or would it be better to develop deep expertise in a few things? Areas where even smart AI with reasoning still can get tripped up.

Trying to broaden your base of expertise seems like it’s always a good idea, but when AI can slurp the whole internet in a single gulp, maybe it isn’t the best allocation of your limited human training cycles.

j_maffe  5 months ago

Do you have any specific tips for the last point? I completely agree with it and have set up a fairly robust Obsidian note taking structure that will benefit greatly from an agentic assistant. Do you use specific tools or workframe for this?

  • fallinditch  5 months ago

    What works well for me at the moment is to write 'books' - i.e use ai as a writing assistant for large documents. I do this because the act of compiling the info with ai assistance helps me to assimilate the knowledge. I use a combination of Chatgpt, perplexity and Gemini with notebook LM - to merge responses from separate LLMs, provide critical feedback on a response, or a chunk of writing, etc.

    This is a really accessible setup and is great for my current needs. Taking it to the next stage with agentic assistants is something I'm only just starting out on. I'm looking at WilmerAI [1] for routing ai workflows and Hoarder [2] to automatically ingest and categorize bookmarks, docs and RSS feed content into a local RAG.

    [1] https://github.com/SomeOddCodeGuy/WilmerAI

    [2] https://hoarder.app/