FAST-NUCES Lahore. Developing at the intersection of low-level systems programming, machine learning, and interactive graphics. Building scalable and efficient digital architectures.
Systems engineering, high-performance computing, and full-stack architecture. Hover over nodes in the interactive constellation network to map system connection pathways and expertise.
Waste routing, logistics tracking, and NGO-focused cooperation platform using Node.js, JS, and secure SQL databases.
Three-lane racing arcade written from scratch in raw 16-bit Assembly with direct BIOS hardware rendering.
A high-performance, localized social network desktop clone built from scratch in C++ and SFML with custom glassmorphism rendering.
Classic 2D retro space shooter built in C++ that renders directly inside the Windows console client area using double-buffered GDI.
End-to-end NLP & machine learning pipeline deploying statistical baselines, BiLSTM networks, and fine-tuned DistilBERT transformers.
A C++ system strictly enforcing SOLID principles and behavioral patterns. Utilizes dynamic runtime polymorphism for role actions, and decouples the Data Access Layer using template-driven functional callbacks to completely isolate file-handling invariants.
Low-level operating systems optimization, structural software architecture design, and commercial feasibility studies.
Academic foundation at FAST-NUCES, Beaconhouse schooling, certifications, and advanced hardware environment optimizations.
BS Computer Science core modules and operating systems research.
Graduated A-levels focusing on maths, computer science and physics.
Get in touch for collaboration opportunities, project inquiries, or any general questions. I look forward to connecting with you.
Developed using Node.js and custom SQL databases, EcoTrack is a waste routing, tracking, and NGO-focused system. It solves logistics for waste collection and streamlines donor-NGO cooperation.
A three-lane arcade racing game developed entirely in 16-bit x86 Assembly language. Runs in a DOS box emulator, rendering graphics directly via bios interrupts and manual pixel mapping.
The Code Evidence (CsvHelper.h): Your readAll method uses a generic template T paired with a lambda parser function: std::function<T(std::stringstream&)> parser.
Design Principle Proof: This perfectly implements the Template Method behavior. The invariant steps (opening files, stream checks, line iteration loops) are fixed within CsvHelper. The variant steps (how a line becomes a Staff, Customer, or Vehicle object) are deferred and injected via custom lambdas. This cleanly achieves Open-Closed Principle (OCP) compliance—you can add infinitely many new models to your system without ever modifying the core file-parsing engine.
The Code Evidence (Staff.h): You designed a clean polymorphic hierarchy using an abstract base class (Staff) equipped with a pure virtual method virtual std::string getRole() const = 0; and a virtual default destructor.
Design Principle Proof: This satisfies the Liskov Substitution Principle (LSP) and Interface Segregation. By utilizing dynamic runtime binding through overrides (Manager, Salesman, Surveyor), high-level authentication services can interface entirely with the Staff abstraction without needing to couple themselves to concrete, low-level roles at compile time.
The Code Evidence (Staff.h): Data members (id, name, username, password) are strictly hidden behind protected boundaries, exposed uniformly via constant reference getters (const std::string& getName() const) to prevent accidental mutability or data leakage.
Design Principle Proof: It respects the Single Responsibility Principle (SRP) by making objects completely autonomous; the serialize() routine manages string conversion logic inside the entity block itself, maintaining structural encapsulation.
PakistanHub is a localized, interactive social network clone built entirely in Object-Oriented C++17 on top of the Simple and Fast Multimedia Library (SFML 2.5+). It brings modern web-like design aesthetics, including custom glassmorphic rendering and smooth layout transitions, to a native desktop environment.
A classic 2D retro-style Space Shooter game built entirely in C++ that renders graphics directly inside the Windows console client area. By bypassing standard terminal text limits and utilizing Windows GDI (Graphics Device Interface), the project achieves smooth frame rates and double-buffered graphics inside a CLI environment.
Engineered an end-to-end machine learning and natural language processing pipeline in Python to classify crime narratives and detect deceptive text across three distinct architectural paradigms. Deployed a unified inference gateway capable of serving statistical, deep recurrent, and transformer-based models under strict resource constraints.
generate_data.py/generate_deception_data.py) to systematically generate domain-specific statement variations, expanding data density and improving model validation benchmarks against sparse training sets.Deep-diving into low-level engineering, memory management, and system-level performance. Leveraging structural software patterns and rigorous process scheduling to build optimized, high-efficiency software architectures.
Conducted extensive market feasibility and financial planning research for sports facility setups.