Hello, world — I am
I'm Rafay — an AI & Machine Learning Developer crafting intelligent systems with Machine LearningDeep LearningNLPPython that turn raw data into real decisions.
BASED IN ISLAMABAD, PAKISTAN · WORKING WORLDWIDE
Every developer has an origin story. Mine starts with a falling block, a blank screen — and the question "what if the machine could learn?"
Chapter 01 — The First BuildBefore the models and the metrics, there was a game. As a new BS Artificial Intelligence student at Air University, Islamabad, I built a complete Tetris clone in C++ — game loop, scoring, levels, all of it. It was the first time I took a program from an empty file to something people could actually play, and I was hooked on finishing things.
Chapter 02 — Machines That LearnThen came the real fascination: software that improves itself. I moved from writing rules to training models — first prediction models with scikit-learn, then recommendation engines, classifiers and clustering systems on real datasets. An 8,790-title Netflix dataset taught me more about messy, real-world data than any textbook could.
Chapter 03 — Learning in the FieldThrough three Machine Learning internships, I worked as a Machine Learning Intern on production-style problems: cleaning data, engineering features, comparing algorithms and defending the results. My best model reached 97.5% accuracy — and the habit of measuring everything stayed with me.
Chapter 04 — The Next PageToday I'm going deeper into Deep Learning and NLP, shipping open-source projects and a live ML demo anyone can try. The story is still being written — and I'm open to remote AI/ML internships worldwide for the next chapter.
From a first C++ game to production-style machine learning — the milestones that shaped how I build.
Current Focus · Open Source
Now focused on neural networks and natural language processing, building in public on GitHub — and open to global remote AI/ML internships.
Machine Learning Internships · Remote
Completed three ML internships working on real datasets end to end — data cleaning, feature engineering, model training, evaluation and reporting to professional standards.
GitHub · GitHub Pages
Put the Netflix recommendation engine live on the web for anyone to try, and published six open-source projects with documented code and results.
Self-Directed · scikit-learn
Built my first prediction models — classification, regression and clustering — and discovered the craft of turning raw data into decisions.
C++ · Game Development
Shipped a complete Tetris clone from scratch: full game loop, line clearing, scoring and progressive levels. My first finished, polished program.
Air University · Islamabad
Began my degree in Artificial Intelligence — and decided from day one to build real things alongside the coursework, not just study theory.
Real models, real datasets, real numbers. Each one is open source — dive into the code.
An SVM-powered system that reads handwritten characters and classifies them with 97.5% accuracy — my highest-performing model. Careful preprocessing and feature extraction keep it robust across wildly different handwriting styles.
A full machine learning suite on 8,790 Netflix titles: a TF-IDF recommendation engine, content-type prediction at 91.9% accuracy, rating classification, and K-Means clustering that surfaces hidden content segments.
A Random Forest credit-risk scorer that predicts loan default risk with 75% accuracy and a strong F1 of 0.83 — tuned to catch risky applicants without unfairly rejecting the good ones.
Early diabetes risk prediction comparing Random Forest and SVM classifiers on clinical data — 74% accuracy, with feature analysis revealing which health indicators matter most.
The classic block-stacking game, built entirely from scratch in C++ — full game loop, line clearing, scoring and progressive levels. The project where my programming journey really began.
Not a list of logos — the tools I've actually shipped with, measured by the work they've produced.
The workflow behind the numbers: data cleaning → feature engineering → model selection → honest evaluation. Every skill above earned its percentage in a shipped project or an internship task — nothing decorative.
I'm open to remote AI/ML internships worldwide, freelance ML projects and collaborations. My inbox is always open — I usually reply within a day.