Let’s not begin with tools. Not with your GitHub streak, or the anxious rush of unread PR reviews, or even the heroic fantasy of fixing production in a midnight haze of broken deployments and cold pizza.
Not even with code, really.
Let’s begin with the thing no one tells you when you start: that your job, your actual job, is to learn.
Not just now. Not as a phase. Not as something you do in a bootcamp or during your first year or when preparing for an interview.
Always.
The great missdirection - disguised as advice, perpetuated by those who speak in metrics and milestone decks - is that the engineer is a builder of software. A shaper of abstractions. A solver of problems.
All true, but all beside the point.
Because behind all of it, underneath the frameworks and the tools and the open source and the systems design and the design docs and the blessed and cursed org charts, there is one persistent fact.
You are paid to deliver value. And that value depletes every day.
Not because you’re failing. But because the ground shifts. Slowly. Constantly. Invisibly. The language evolves. The infrastructure patterns change. The business priorities tilt slightly. A tool you built six months ago is now being rewritten by someone younger, someone hungrier, someone who doesn’t yet feel tired in the way you do.
And here’s the hard part: it’s no one’s fault. It’s just what the job is.
So, you learn. Or you drift. You evolve. Or you fade.
This is not a threat. It’s a rhythm. One that begins again every morning when you sit down, blink at your screen, and remember that the very thing that made you good at this - the curiosity, the stubbornness, the restlessness - is the only thing keeping you in it.
The Perpetual Motion of Learning
You might be tempted, at some point, to view this cycle of learning as a burden. You might see it as an anxiety machine, constantly ticking in the background, whispering that you're always behind, that you don't know enough, that someone somewhere is moving faster.
But this is the wrong story. And it’s a dangerous one.
Because this - this constant, cyclical, iterative becoming - is not punishment. It is the work. It is the structure. It is the source.
There is a quiet joy to be found in this kind of work, the kind of joy that doesn't shout but stays. The kind that builds something inside you. Not an empire. A motor.
This is what it means to be a perpetual learning machine.
Not a productivity engine. Not an automation wizard. Not a careerist with a Kanban board full of job titles and compensation bands. But a person who learns because they cannot imagine not learning.
The Cause and the Consequence
You are here because of your own momentum.
Your curiosity brought you here. You pulled at threads. You opened tabs. You stayed up late watching talks about things you didn’t understand, not because you were told to, but because something inside you needed to know.
That curiosity is not incidental. It is the material. You didn’t become an engineer and then become curious. You became an engineer because you already were.
And yet, at some point - often quietly - we begin to treat the job as a destination. We think we’ve arrived. We accumulate years, and those years start to sound like proof. As if tenure could replace motion. As if stability could replace growth.
But the field doesn’t care. The ground still moves.
And you, if you’re honest, still feel it. That need to know. To build. To revise.
So don’t run from it. Don’t pathologize the movement. Don’t resent the pull. It’s not a glitch. It’s the engine.
The job is not to become. The job is to keep becoming.
You are the cause of your own continuity. You are the source of your own fuel. You are the one thing that ensures you remain sharp, relevant, useful, creative.
You are, if you let yourself be, a perpetual learning machine.
You are a tinkerer, and tinkeres never stop.
On plumbers and prophets
When Geoffrey Hinton was asked what he would tell people about the future of their careers in an era where machines write poems and pass the bar exam said: train to be a plumber.
He meant it.
Instead of answering with grand predictions or VC-friendly slogans, he pointed to the ground. To pipes. To homes. To work that resists automation not because it’s dumb, but because it’s real.
Plumbing, he suggested, is safer than software.
You fix what is broken. The materials are consistent. The tools do not reinvent themselves every third quarter. A good plumber in 1995 is a good plumber in 2025.
And maybe you laugh. Or maybe you pause.
Because here you are, debugging a stack of abstractions so distant from physical reality you no longer remember what any of them actually do. React components that wrap other components. Containers running in orchestrated clusters. Code that you wrote and forgot and that someone else will soon forget too. A job made almost entirely of shifting sand.
You cannot be still in this job. The minute you pause, the system changes around you. The framework updates. The tool becomes obsolete. The architecture you once defended is now being decommissioned.
So what’s the alternative?
You move.
Embracing Perpetual Motion
Let’s talk about how to live in the job. Not survive it. Not sprint through it and collapse into the next one. But live in it - year after year - without burning out or calcifying. If you’re going to be a perpetual learning machine, you can’t run on panic. You’ll need something slower. Deeper. Repeatable.
You’ll need a system that protects your attention like a precious resource. Because it is.
The Battery Isn’t a Metaphor
Let’s make one thing disorientingly, unequivocally clear:
Everything costs energy.
Not in the armchair‑philosophy, productivity‑hack sense of the word, but in the basal, metabolic, physiological sense. Your brain, that lush jungle of spiking neurons and leaky abstractions, runs on glucose. And the energy it takes to decide what to do, to keep a line of thought intact, to imagine something that does not yet exist, to learn - it’s real. It’s chemical. It depletes you.
There’s no separate energy silo for entertainment or “fun.” No moral distinction between what drains you “productively” and what drains you pleasurably. Scroll through Instagram for forty minutes and you’re not recharging - you’re spending. Try to kill your friends for two hours of Jiu‑Jitsu? Still spending. Wrestling, walking, talking, reading, watching, listening. The machinery is always on. The cost is always logged.
What we call “mental fatigue” is not a mood. It’s not a vibe. It’s a measurable state of depleted neurological capacity. There are fMRI studies that show it.A fatigued brain avoids hard tasks even when the rewards increase. The brain isn’t lazy - it’s economical. If the circuits responsible for executive function and attention are tired, they downregulate. They say no. They click the video, they open the fridge, they bounce off the learning loop because the loop, quite literally, costs too much.
And none of this is an argument for abstinence. Quite the opposite. Wrestle. Scroll. Run. Let your brain light up in ways that keep you whole and alive. Joy, after all, is one of the few forces that justifies the burn. But the ledger is real. The battery doesn’t care what the task was. It only knows how much was used.
Just remember they are not free. Do not pretend that your energy is infinite just because your curiosity is. Do not make the mistake of mistaking stimulation for surplus.
The battery depletes. Daily. Quietly.
Be smart and intentional about how you spend it.
Habit Stacking and the Illusion of Discipline
People think they need more willpower. More grit. More hustle. But most days, you don’t need a heroic effort. You need a hinge.
Something small. Something already moving.
The mind - especially a tired, overstimulated one - defaults to inertia. It doesn’t leap. It slides. And if it can slide from one familiar thing into another without breaking stride, it will. That’s where the habit lives.
This is why brushing your teeth after eating isn’t discipline. It’s choreography. One follows the other because that’s the dance your brain learned. And once it’s learned, it’s cheaper. Less cognitive load. Less friction.
So if you want to read more, leave the book out - open, not closed. If you want to write more, end your writing session mid-sentence, so you’re already in the middle of something the next time you sit down. If you want to study, tether it to something you already do without thinking. First coffee. Last scroll. Post-shower.
Brazilians shower more than any other people on Earth. That’s three anchor points a day for brushing one’s teeth and brushing is an excellent proxy for flossing. That's how you can transform an habit of showering into a system of dental care.
You don’t need to invent a routine from scratch. You need to attach your new motion to an old rhythm.
Habits aren’t made from scratch. They’re smuggled in.
Friction and the Quiet Death of Momentum
You don’t stall because the task is hard. You stall because you had to open three tabs to find it. Because the doc you need is in a folder inside a folder inside a system that still calls itself “intuitive.” Because you forgot whether the video was at 27:43 or 28:17 and when you went looking for it, your brain caught a glimpse of a thumbnail labeled “Things You’re Probably Doing Wrong in Rust,” and that was it. Game over.
Friction doesn’t scream. It doesn’t punch. It whispers.
It’s a barely perceptible delay, a micro-confusion, an “I’ll do it right after…” that never resolves. It’s not the size of the task that ruins you. It’s the presence of just enough drag that your mind - a beautiful, lazy, cautious animal - decides the effort-to-value ratio has tipped. And in that moment, the loop breaks. Not catastrophically. Just enough to lose the beat.Just enough that you check the news. Or Slack. Or your nails.
And that’s the part to understand: the system - your system - is not made of willpower. It’s made of loops. Smooth ones. Feedback ones. Ones with as little hesitation as possible.
The goal isn’t to summon more energy. It’s to lose less of it between intention and action.
To know that if the book is already open on the table, the page marked, the note beside it waiting, you’re already studying. That if the gym bag is packed, the toothbrush in your bag, the draft tab open from last night - then half the decision has already been made. You’ve shortened the distance between wanting and doing.
It’s not about hacks. It’s about not giving your brain an excuse to think. Because friction - just like entropy - wins quietly. And learning, like all forms of becoming, needs silence, not just in sound,
but in resistance.
Don’t build a stronger will. Build a smoother path.
For studying, the best way to do it is crafting a plan.
Building a System That Carries You
This is the part where we stop talking about theory and start showing. But before we get tactical, let's lay one thing bare: you can’t rely on memory, motivation, or inspiration. You need to outsource those to a system - one that doesn’t need daily negotiation.
That’s the game: reduce friction, reduce variance, reduce the psychic tax of deciding what to do next. A good plan doesn’t just show you the way. It quiets the noise in your head long enough for you to actually go.
Treat it like a personal university. A curriculum. Not a to-do list. Not vibes. Not “whatever course feels cool this week.” A curriculum is different. It presupposes that you’re serious. That there’s a body of knowledge to move through. That you are building, not wandering.
Let me show you how I do it — not because it’s universal, but because it’s concrete. You’ll tweak it. You should. But steal the architecture.
Step 1: Decide What to Learn
Pick 2 to 3 themes for the year. Not ten. Not “AI and DevOps and Management and Rust and Kubernetes and Cooking.” Time is limited. Focus forces depth. Be specific.
In my case, 2026 is about deepening expertise across AI systems, vision models, and infrastructure — the kind of work that demands both architectural thinking and production-level execution.
- AI models and Large Language Models
- Computer Vision
- Systems & Platform Design
- Staff-level behavioral skills
Step 2: Build a Curriculum
You’re not just collecting resources. You’re sequencing them. Choosing what to study, what to skip, and in what order. Courses, books, exercises — all stacked. This is what turns a random walk into a path.
Here’s a real excerpt from my curriculum.md:
| ⏺ | Quarter | Focus |
| --- | ------------------------- | --------------------------------------------------------------- |
| Q1 | Production LLM Systems | Multi-tenant LLM platform with RAG + monitoring |
| Q2 | Computer Vision at Scale | Real-time CV inference API + performance optimization framework |
| Q3 | AI Platform Engineering | Unified ML platform + comprehensive architecture documentation |
| Q4 | Enterprise AI Integration | Production multi-agent system + cross-team adoption metrics |
It’s not overwhelming. It’s clear. Each quarter has a theme and a thing you’re shipping. No ambiguity. No rabbit holes.
Step 3: Create Pools, Not Chaos
Each theme gets its own pool of resources: courses, books, exercises. I don’t pretend I’ll finish them all. But I know where to pull from. So when it’s 7:00pm on a Tuesday and I’ve got 90 minutes, I’m not asking “what should I do?” I already decided that weeks ago.
Here’s a snapshot of my LLM pool:
### AI Applications (Making ML Practical)
- "Hands-On Machine Learning with Scikit-Learn, Keras & TensorFlow" book
- Fast.ai Practical Deep Learning course
- MLflow documentation and tutorials
- Kaggle Learn: Intro to Machine Learning
- "Data Science for Business" book
- Feature Engineering for Machine Learning course
- "Interpretable Machine Learning" book (Christoph Molnar)
- "Machine Learning Design Patterns" book
Simple. Trackable. No mystery.
Step 4: Weekly Routine (The Beat That Keeps You Moving)
This is the core of it. Everything feeds into a weekly schedule that’s stable but flexible. Like this: • Mon–Tue: Study • Wed–Fri: Project work • Weekend: Writing + Reflection
# Weekly Practice Routine
**Total Time**: 15 hours/week
---
## Weekly Schedule Template
### Monday (2 hours)
** Study Time**
- [ ] **90 min**: Study from current quarter's primary pool
- [ ] **30 min**: Read from Staff Engineering Pool (throughout year)
- Focus: Career development, technical leadership, communication
**Current Resource:**
- Book/Course: _[Write current resource here]_
- Chapter/Section: _[Track progress]_
---
### Tuesday (2 hours)
** Study
- [ ] **2 hours**: Continue study from current quarter's pool
- Same pools as Monday
- Alternate between theoretical (courses/books) and practical (tutorials/docs)
---
### Wednesday (3 hours)
** Project Work: Deep Focus**
- [ ] **3 hours**: Work on current quarter's project
- [ ] **15 min**: Update project README/documentation as you go
**Current Project:**
- Name: _[Current project name]_
- This Week's Goal: _[Specific milestone]_
- Blockers: _[Any issues to resolve]_
---
### Thursday (3 hours)
** Project Work: Build & Iterate**
- [ ] **3 hours**: Continue project work
- Focus on implementation
- If stuck, research solutions (counts toward project time)
- Test as you build
- [ ] **Checkpoint**: Can you demo something today?
- Yes: Record a quick demo or screenshot
- No: Identify what's blocking you
**Progress Notes:**
- What I built today: _[Brief notes]_
- Challenges: _[What was hard?]_
- Learnings: _[What did I discover?]_
---
### Friday (2 hours)
** Project Work + Polish**
- [ ] **1.5 hours**: Project work (wrapping up week's milestone)
- [ ] **30 min**: Code quality & best practices
- Refactor messy code
- Add tests for critical functionality
- Review your own code (pretend you're reviewing someone else's PR)
- Update comments and documentation
**Weekly Milestone:**
- [ ] Did I achieve this week's project goal?
- [ ] What's the goal for next week?
---
### Weekend (2 hours)
** Writing & Reflection**
#### Saturday or Sunday (2 hours)
- [ ] **1.5 hours**: Technical Writing
- **Options**:
- Write/continue blog post
- Draft RFC or design document
- Update project documentation
- Write ADR (Architectural Decision Record)
- Contribute to open source docs
- **Monthly cadence**:
- Blog post (every month)
- RFC/Design doc (alternating months, or as needed for projects)
- [ ] **30 min**: Weekly Review & Planning
- Review what you learned this week
- Update curriculum progress (check off completed resources)
- Plan next week's focus
- Update quarterly progress tracker
**This Week's Writing:**
- Topic: _[What are you writing about?]_
- Progress: _[Draft started / halfway / ready to publish]_
It’s mechanical, yes. That’s the point. Friction kills. Routines save.
Step 5: Track It or It Didn’t Happen
I log hours. I track progress. I do monthly reviews. Not because I’m Type A — as if those kind of artificial boxes made any sense in first place — but because I am dumb and lazy. I forget otherwise. And tracking creates the loop: intention → action → feedback → refinement.
This is how you sustain learning over 12 months. Not with willpower. With systems.
The Loop Is the Point
There’s a kind of symmetry in the arc.
You started here because something tugged at you. A curiosity. A puzzle. A sense that you could build something — or at least, understand it. You weren’t chasing titles or optimizing for output. You were chasing a thread.
Then it became a job.
And slowly, that initial spark got buried under tools and meetings and tickets and all the rest of it. You still learned — you had to — but learning became background noise. Maintenance. An obligation to stay “relevant.”
But maybe — just maybe — this is the moment you return to it. Not as nostalgia. Not as a side project. But as the foundation.
Because that instinct — to learn, to build, to revise — it was never just a phase. It wasn’t a stepping stone. It was the engine. It still is.
This isn’t about climbing. It’s about becoming — in the Deleuzian sense. Not aiming at a fixed goal, but remaining in motion. Sustaining the rhythm of transformation, without needing to resolve into a title, a tier, a shape.
To become, in this way, is to resist calcification. To keep your attention supple. To let your practice evolve with your questions.
That’s why we build systems. Not to optimize. Not to maximize. But to stay close to the original rhythm. To hold open a space where your curiosity can still breathe. Where learning isn’t just permitted, but expected. Protected.
So build the curriculum. Track the work. Make the loops tight. Not because it makes you more productive, but because it keeps you connected — to the person who started all this.
You are still that person.