Grade 12 · Aspiring CS & AI student

Python Developer &
Machine Learning Developer

Building machine learning systems and AI applications with Python — from data preprocessing and model training to deployment.

View My Work Download CV
Portrait of Rakan Tahineh
Rakan Tahineh Python · ML · AI
01 — About

A bit about me

I'm Rakan, a Python developer with experience in web development and a strong focus on Machine Learning and Artificial Intelligence. I began my journey through web applications, which helped me build strong programming fundamentals and an understanding of how real-world systems are structured. Over time, my interests shifted toward AI and ML, where I now focus on building data-driven systems and predictive models.

What motivates me is solving real-world problems using data and computation. I'm particularly interested in how machine learning models learn patterns from complex datasets and turn them into useful, practical applications. I enjoy end-to-end projects spanning data preprocessing, feature engineering, model training, evaluation, and deployment.

Currently, I'm strengthening my machine learning foundations and expanding my portfolio through applied projects such as predictive modeling systems and AI-based applications — while experimenting with different models to understand their behavior in real-world scenarios.

My goal

To pursue advanced studies in Artificial Intelligence and contribute to the development of practical AI systems that solve real-world challenges.

02 — Projects

Selected work

Applied machine learning and full-stack projects — more on the way.

Gaddum Stone Age collection — cream and navy t-shirts against a desert landscape
Django PostgreSQL Stripe RAG Docker

Gaddum — Modern Essentials

The full e-commerce platform behind my own UAE t-shirt brand — designed, built and deployed end to end. Browsing and search, cart, promo codes, Stripe checkout, order tracking with delivery updates, customer accounts with wishlists and saved addresses, and an admin dashboard the shop runs on day to day without touching code.

  • RAG shopping assistant answers only from a purpose-built index of the shop's own content via PostgreSQL full-text search — and says so when the answer isn't there
  • Privacy by construction: order lookups are bound to the signed-in customer server-side, and stock counts and promo codes are never indexed at all
  • Inventory can't oversell — stock is claimed inside a transaction with row-level locking, so the database decides who gets the last item
  • Prices frozen at purchase and promo limits enforced atomically, so editing a product never rewrites order history
Livegaddum.store
278Automated tests
RAGShop assistant
StripePayments
DockerDeployment
Visit the Store Get in Touch
UFC Fight Outcome Predictor showing win probability, odds, and tale of the tape
Machine Learning Flask Web Scraping MLOps

UFC Fight Outcome Predictor

An end-to-end ML system that predicts UFC fight outcomes — win probability, betting odds, and finish method — for any hypothetical matchup. It scrapes ufcstats.com, builds leakage-audited point-in-time features (including a custom Elo rating), benchmarks Logistic Regression against XGBoost, and serves a calibrated model through a live Flask app that retrains itself weekly via GitHub Actions.

  • Calibrated Logistic Regression matches XGBoost at the honest ~62% ceiling for UFC prediction
  • Two leakage traps engineered away: point-in-time features only, and symmetric corner-swapped inputs
  • Fully automated: weekly cloud retrain scrapes new fights, re-benchmarks, and redeploys the demo
2,700+Fighters covered
~62%Accuracy (honest)
WeeklyAuto-retrain
LivePublic demo
Try the Live Demo View Repository
UAE Car Price Predictor showing an estimated price and range
Machine Learning Flask Scikit-learn

UAE Car Price Predictor

A machine learning web app that estimates used-car prices in the UAE market. It trains Linear Regression and Random Forest models, then blends them into a weighted ensemble for a more reliable estimate — with a clean Flask interface returning an instant price plus a likely range.

  • Ensemble MAE of ~39,700 AED — lower error than either model alone
  • Trained on 7,200+ real listings across 100 brands & 716 models
  • End-to-end pipeline: cleaning, feature engineering, training, deployment
7,200+Listings analyzed
100Brands
716Models
EnsembleML pipeline
FlaskDeployment
View Repository Get in Touch

More projects coming soon — I'm actively building.

03 — Now

Currently building & learning

Currently Building

  • AI Club Launching Grade 12
  • Road to AI — YouTube Channel Live
  • ML Specialization — Coursera In progress

Currently Learning

TensorFlow Deep Learning Neural Networks
04 — Skills

Tools & technologies

ML / AI Frameworks

Scikit-learn TensorFlow learning

Languages

Python JavaScript SQL

Web / Backend

Flask Django HTML / CSS

Data Tools

Pandas NumPy Matplotlib Jupyter

Other

Git / GitHub Model deployment ML web integration
05 — Contact

Get in touch

Questions about my work, or just want to connect? I'd love to hear from you.