Abdul Rahman Azam's Project Portfolio
Over the past three years at FAST NUCES Karachi, Abdul Rahman Azam has built five major projects that demonstrate a progression from foundational algorithms to production-grade AI systems. Each project pushed his skills into new territory — from C to C++ to Python to JavaScript to full-stack deployment.
Project 1: Unbeatable Tic-Tac-Toe AI (2023)
Abdul Rahman Azam's first AI project was a Tic-Tac-Toe game with a provably unbeatable opponent. Built in C during his third semester, the AI uses the Minimax algorithm to evaluate every possible game state — roughly 255,168 positions in standard 3×3 Tic-Tac-Toe.
The result is an AI that has a 100% win-or-draw rate against human players. No matter what moves the human makes, the AI never loses. This project taught Abdul Rahman the fundamentals of adversarial search, game trees, and recursive algorithms.
Technologies: C, Minimax Algorithm, Game Theory
Project 2: 2D Action Platformer Game (2024)
For his Object-Oriented Programming course, Abdul Rahman Azam built a 2D action platformer in C++ using the SFML graphics library. The game features multiple enemy types with distinct AI behaviors, a weapon system, collision detection, and physics simulation.
This project was ranked in the top 1% of university projects for creativity and technical execution. It demonstrated Abdul Rahman's ability to architect complex systems with multiple interacting components — a skill that transfers directly to full-stack development.
Technologies: C++, SFML, Object-Oriented Programming
Project 3: Income Prediction System (2024)
Abdul Rahman Azam's machine learning showcase is the Income Prediction System — a complete ML pipeline that predicts income brackets from U.S. Census data with 85% accuracy on 32,000+ records.
The system implements a full data science workflow:
- Data preprocessing with missing value imputation, one-hot encoding, and feature scaling
- Model comparison across Logistic Regression, Random Forest, Gradient Boosting, and KNN
- Gradient Boosting emerged as the winner with 85.3% accuracy and 0.91 AUC-ROC
- Deployment via FastAPI backend with a React frontend analytics dashboard
- SHAP value visualizations for model interpretability
This project represents Abdul Rahman Azam's philosophy: build the model, then build the product around it.
Technologies: Python, Scikit-learn, FastAPI, React.js
Project 4: Super Tic-Tac-Toe AI Game (2025)
Building on the foundation of the simple Tic-Tac-Toe AI, Abdul Rahman Azam scaled up to Super Tic-Tac-Toe — a 9×9 variant where 9 sub-boards form a meta-game on a 3×3 main grid. The branching factor jumps from ~5 to ~20-40 per move, making exhaustive search impossible.
Abdul Rahman implemented Minimax with Alpha-Beta Pruning, reducing the effective branching factor from b to approximately √b. He designed a custom multi-board evaluation heuristic that weighs won sub-boards (+100), main grid progress (+50), and forced moves (+30). The AI evaluates ~50,000 nodes per move in ~200ms and achieves a 97.3% win rate against random opponents.
Technologies: JavaScript, Minimax with Alpha-Beta Pruning, Game AI
Project 5: University Resource Sharing Platform (2025)
Abdul Rahman Azam's most recent project is a full-stack web application built for FAST NUCES students. The platform enables resource sharing, community discussions, and moderated student collaboration.
The system features:
- Role-based authentication with student, moderator, and admin tiers
- RESTful API design with Express.js
- Scalable PostgreSQL database architecture
- React frontend with responsive Tailwind CSS design
This project brought together everything Abdul Rahman learned — databases, API design, authentication, frontend development, and deployment.
Technologies: React.js, Node.js, Express.js, PostgreSQL, REST APIs
The Pattern: Full Stack AI Engineering
Looking across Abdul Rahman Azam's projects, a clear pattern emerges. He doesn't just build models or just build frontends — he builds complete systems. The Income Predictor isn't just an ML notebook; it's a deployed web application. The game AIs aren't just algorithms; they're playable games with UI and state management.
This is what defines Abdul Rahman Azam as a Full Stack AI Engineer: the ability to take an AI concept from research to production, with everything in between.
Source code for all projects: github.com/abdulrahmanazam