AI and LLMs
What these tools are genuinely good at, what they are not, and how to run them yourself.
I use these tools daily and I am neither evangelical nor dismissive about them. Mostly I am interested in where the line sits between useful and confidently wrong, and in running as much of it as possible on my own hardware.
5 posts on this.
Start here
Creation Is Art, Debugging Is Science
Why models are so much better at finding your bug than designing your system.
Everything on this topic
- Creation Is Art, Debugging Is Science Why models are so much better at finding your bug than designing your system.
- Copilot Is Training Wheels You Should Eventually Remove Useful while you are learning, and worth putting down once you have.
- Building Custom GPTs for Your Workflow Building a narrow assistant that knows your workflow rather than everything.
- Building a RAG Pipeline with LangChain Getting a model to answer from your documents instead of inventing things.
- AI Won't Replace Developers, But It Will Replace Developers Who Don't Use AI The tooling is not coming for your job, but the people using it might be.
Available for Node.js and TypeScript contracts
Senior software engineer and Node.js contractor, 15+ years across fintech, betting and gaming, media, and e-commerce. If this is the kind of work you need doing, my full history is on the CV.