DEV Community: Fachri

Hi folks! In case you guys havent heard of this. Support will be awesome. Spread the words!

Fachri — Mon, 20 Oct 2025 11:36:49 +0000

Introducing a new learning platform for Language Model.

Fachri ・ Sep 18

Why I Keep Coming Back to Go (After Trying Almost Everything Else)

Fachri — Fri, 19 Sep 2025 11:28:04 +0000

I’ve tried a lot of languages in years of years building products.

Back when Node.js came out, I was all in. The event loop was mind-blowing, JavaScript everywhere was appealing, and the community energy was wild. I shipped real stuff with it. But as things got more complex, I started to feel the weight of callback hell, package chaos, and eventually TypeScript debates. It was fun, but it didn’t scale with the same simplicity.

Rails? I loved how fast I could build MVPs. Convention over configuration worked… until the abstraction layers started to leak and it became harder to trace what was actually happening under the hood. But overall, the community is still awesome, and DHH is a role model for many things for software development mind. Respect.

Elixir with Phoenix was next on my list. Super elegant. Lightweight. The actor model was fascinating. But I quickly realized it wasn’t something I could easily hire for, and maintaining it solo didn’t feel sustainable. Phoenix is somewhat using the same method like rails, but yeah it might be suited for more to a real time and heavy application.

Then I explored Rust. I respect it. A lot. But let’s be honest: I’m not building kernel modules every day. I needed something fast to ship with, not something that would slow me down with lifetime annotations just to build a basic API.

That’s when Go started to stick.

I didn’t fall in love with Go because it was cool. I fell in love because it got out of my way. It’s boring in the best possible way. The syntax is so simple I barely have to think about it. There are no clever tricks, no magical metaprogramming. Just clear, type-safe, readable code. *This is what important. *

Go fails fast. If you mistype something or mess up your types, it tells you immediately. That alone saved me from so many stupid bugs. And once it compiles, I’m confident it's going to run — and run fast. There are caveats, but you will know when you are used to it.

I’ve now shipped multiple products using Go. APIs, backend services, CLIs, internal tools. Every single one has felt seamless. The tooling is first-class. The build times are ridiculously fast. Cross-compilation is easy. Deployments are predictable. And maintenance is, honestly, kind of a joy.

The ecosystem is mature too. That’s not a coincidence. Go is the language of infrastructure for a reason. Find the article or videos and understand why golang is exist at the first place.

Sure, Go isn’t perfect. Error handling can feel repetitive. And yeah, generics came late. But in practice? None of that has ever blocked me from building and shipping fast. Go just works. That’s why it’s still my default, not because it’s flashy, but because it helps me stay fast and sane as I scale.

I’m curious what others think.
Have you gone through a similar journey? What’s the language you trust the most when it’s time to ship?

Introducing a new learning platform for Language Model.

Fachri — Thu, 18 Sep 2025 07:20:46 +0000

After mode than decades building stuff (startups, internal tools, side projects), and watching AI go from “weird paper from Google” to “this thing might actually replace my job,” I started getting frustrated with how disconnected most AI content is from actual understanding.

So I made LMPath on several weekeends. It’s not another wrapper for ChatGPT or some generic AI course with surface-level fluff. This is for people who actually want to understand how this shit works.

So without a further a do, without some make up. This is it. I try to be direct of what it is to not waste your time.

What it is

Real explanations (how transformers/GPT actually work, not just “type prompt, get magic”)
Based on the original papers (Attention Is All You Need, GPT-3, RAG, etc.)
Geared for devs — no insane math walls, just enough to get the core ideas

What it’s NOT

It’s not a business guide or “how to start an AI startup”
It’s not a full-blown ML course — fast.ai and others already nailed that

What’s working so far

Dark mode (duh)
Auto-progress tracking (pick up where you left off)
Smart breadcrumbs
Chapter progress bars
Deep linking to share a specific part of a lesson

What’s missing

Only 2 courses live for now (Intro to LMs, RAG systems)
No in-browser coding yet
No advanced stuff like training dynamics or distributed inference
I’m building it solo, so yeah, takes time

What’s coming

Hands-on coding labs (real transformer code, not just theory)
In-lesson Q&A assistant
More courses: fine-tuning, embedding, transformers internals
Practice problems (so you’re not just reading and forgetting)

Why I built it
Because I needed this. I wanted something between “LLMs are magic!” and “here’s a 200-page derivation of self-attention.” I needed to know how things worked so I could build better stuff. Maybe you’re in the same boat.

Anyway, it’s free for now. Please check this out LMPath Feedback is very welcome, especially if something’s confusing for you. That’s where the best fixes come from.