I build things because coursework feels too small. From a real-time ICU monitoring system deployed on AWS to a Sign Language to Urdu translator — every project starts with a real question: what if this actually had to work?
I am a second-year Computer Science student at Information Technology University of the Punjab, Lahore. My journey into serious programming began when coursework started feeling too small — so I started building systems that actually had to work.
Every project I build starts from a real need. My IntelliCare-ICU system started from a question about what a real hospital monitoring backend would require. My Sign Language to Urdu Translator started from a school friendship with hearing-impaired students — we stood in the same room, unable to communicate, and I never forgot that.
We eventually learned some sign language, became friends, and I realised they were exactly like us — just with a different language. That afternoon is why I build.
I prefer understanding things from first principles before reaching for a library. I implemented gradient descent by hand before using PyTorch. I built data structures from scratch before trusting standard implementations. I think that is the only way to really know what you are doing.
Real-time ICU patient monitoring backend in C++17 with 7 custom data structures. Sub-2ms query times, deployed live on AWS EC2.
End-to-end NLP pipeline integrating large language models to help researchers draft and structure academic papers.
Real-time translator combining hand gesture recognition with natural language generation. Built from a real human need.
Full-stack gym platform with 4 role-based dashboards, 15-table PostgreSQL schema. Query times reduced 15x. Deployed on AWS EC2.
Ground-up ML implementations in NumPy. Gradient descent, 5-fold CV, threshold tuning. Benchmarked against Scikit-learn.
Classical CV pipeline. Covariance matrix, eigenspace projection, multi-class expression classification from derived feature vectors.
A PostgreSQL extension in C introducing custom data types and operators. Works at the database internals level below the SQL layer.
Full chess engine in C++ with complete legal move enforcement, check/checkmate detection, save/load state, and interactive Raylib GUI.
Playable side-scroller with sprite animation, collision detection, enemy AI state machine, and clean game-loop architecture.
Reversi game with an AI opponent. Implements game tree search with strategic evaluation for intelligent move selection.
Console-based C++ spreadsheet simulation using a 4-way linked list. Supports insert/delete rows/columns, copy-paste ranges, and cell navigation.
Multi-user file system storing all data in a single .fileVerse container. Socket-based concurrent server with sequential operation processing.
Collaborative mobile app that extracts clothing colours from images and applies colour theory to suggest outfit combinations.
Open to internship opportunities, collaborations, and interesting conversations about AI, systems design, and full-stack development. Feel free to reach out.