The Night I Discovered Machine Learning

It was 2 AM, second year of engineering. I had been doing the usual late-night YouTube wandering — starting from one video and ending up somewhere completely different. Then I clicked on a video about how Netflix recommends movies. I did not sleep properly for the next two weeks.

The Video That Broke My Brain

The video explained, in broad strokes, how Netflix's recommendation system worked. It wasn't deep — just enough to make the idea legible. But the idea itself was staggering: a system that learned your preferences from your behavior, without anyone explicitly programming what "good taste" meant.

Nobody had told the algorithm "action movies are exciting" or "long films are tiring." It had figured that out from millions of user interactions. It had learned.

I remember sitting in the dark, genuinely stunned. Not because it was new information — I'd heard the term "AI" before. But this was the first time the concept had landed with real weight. Machines could find patterns that no human explicitly taught them. They could generalize from examples. They could predict. They could, in some meaningful sense, understand.

The Next Two Weeks

I went down the rabbit hole completely and without apology. At 2 AM I was watching 3Blue1Brown's neural network series. At 4 AM I was reading Wikipedia on gradient descent. I fell asleep with papers half-read on my phone screen. I woke up and kept reading.

I didn't understand most of it yet. But I couldn't stop. The feeling was the same as my first working Java program, but amplified by a thousand — the sense that there was an entire universe of ideas I hadn't yet touched, and that these ideas were both theoretically deep and practically powerful.

"The best obsessions feel less like choices and more like gravity. You don't decide to fall in — you just realize you already have."

From Watching to Doing

After two weeks of watching and reading, I finally opened a code editor. I found a scikit-learn tutorial and trained my first classifier — a decision tree on the Iris dataset. The classic beginner move. But the moment I saw the accuracy score print out — 95% — it hit me that I had made a machine learn from data.

It was terrible code. The model was trivial. The dataset was decades old. None of that mattered. I had crossed from watching to doing.

What the Obsession Looked Like

For the next several months, I read everything I could find. Andrew Ng's ML course. Fast.ai. Papers on arXiv that I barely understood. I built toy projects constantly — a sentiment classifier, a simple image recognizer, a basic chatbot. None were impressive. All were essential.

I also started asking deeper questions about math I'd seen in school — why was the derivative important? What was a vector actually representing? Why did matrix multiplication work that way? For the first time, school mathematics felt urgent and real, not abstract.

How This Night Changed My Career

That discovery night is the direct ancestor of everything that came after: building JARVIS as my first real AI project, winning a hackathon with an AI resume parser, co-authoring an IEEE paper on Bayesian deep learning, building RAG pipelines on AWS, and achieving ~99% LLM token reduction in a live agentic AI pipeline at Idyllic Services.

None of that was planned. It all flowed from one night of genuine curiosity — from asking "how does Netflix know what I want to watch?" and following that question wherever it led.

What I Learned About Learning

That night taught me something about how I work: I need the spark of genuine fascination before I can sustain deep learning. Mandatory coursework had never produced that spark. A 2 AM YouTube video did.

Since then, I've always paid attention to what genuinely fascinates me — not what seems impressive or career-relevant, but what I'd read about at 2 AM when nobody's watching. That's been my most reliable compass for what to learn next.

If you're early in your programming journey and haven't found your 2 AM topic yet — keep looking. It exists. And when you find it, everything changes.