LEARNING IN PUBLIC

Code, AI, and other expensive rabbit holes.

I'm David Abraham, a software developer who works mainly with PHP, Laravel, Python, databases, APIs, and the occasional ancient system that looks like it was assembled during a thunderstorm. PeakFlow is where I share what I'm learning, building, testing, breaking, fixing, and figuring out.

What I'm digging into

Lately I've been spending a lot of time on AI — not the shiny keynote version, the practical one. The kind where you install things, break things, read confusing error messages, fix one problem, create three more, and eventually learn something useful.

Large language models Local AI models Diffusion models Machine learning AI-assisted dev Python automation Laravel & PHP Rust Developer tooling System mapping

A workshop, not a showroom

🛠️

The process, not just the polish

Not only finished, shiny projects. Also what worked, what failed, what confused me, what finally clicked, and what I'd do differently next time.

🔬

Grounded, not hype

How can local AI actually be useful? What do LLMs do well, and where do they fall apart? Real questions, from the perspective of a working developer.


Notes from the bench

  • LLM experiments & running AI locally
  • Diffusion & image generation
  • Machine learning learning notes
  • Laravel & PHP tips and gotchas
  • Python scripts & AI workflows
  • Rust learning notes
  • Developer tools worth stealing
  • Debugging stories & project writeups

Some posts will be polished tutorials. Some will be project notes. Some will be “here is the thing that wasted my afternoon so maybe it doesn't have to waste yours.”


Learning in public, one broken install at a time

I'm a software developer with experience in PHP, Laravel, Python, databases, APIs, and real-world business applications — especially the kind of work where the job isn't just writing new code, but understanding existing systems, debugging strange issues, connecting tools together, and making complicated things a little less cursed.

I'm also learning Rust, partly because I like the idea of building fast, reliable command-line tools, and partly because apparently I enjoy being humbled by a compiler with standards.

PeakFlow is where I'll share that process: projects I'm working on, experiments with local AI, notes on LLMs and diffusion models, Python workflows, Rust notes, debugging stories, automation ideas, and practical tricks I pick up along the way.

The main idea is simple: keep learning, keep building, and document the useful parts. There will be tools on the table, half-built ideas, weird experiments, occasional smoke, and hopefully a few things worth stealing for your own projects.

— David