Mobile Cricket DRS System
Flutter, Python, Flask
Project Overview
Developed a mobile-based Cricket Decision Review System (DRS) to assist umpires
in making accurate decisions during matches, leveraging computer vision and
real-time video analysis.
Key Contributions
- Worked with 30 developers across 6 teams to build an automated Cricket Decision
Review System
- Built an efficient Flutter app for real-time video capture and review, reducing
manual decision delays by 80%
- Helped integrate and optimize a 6-module Python backend (tracking, edge detection,
trajectory, decision logic), delivering results in < 10 seconds
- Implemented efficient video processing, frame extraction and analysis algorithms to
enhance decision accuracy
Technologies Used
Flutter
Python
Flask
Authentica - AI Content Detection System
Python, Google Colab, TensorFlow, Kaggle
Project Overview
Developed an AI-powered content detection system to identify AI-generated text and
images utilizing SOTA deep learning technologies, enhancing content authenticity
verification.
Key Contributions
- Utilized a Deep Learning detection system for identifying AI-generated content using
embedding-based analysis
- Implemented modular inference pipelines enabling experimentation across detection
strategies and modalities
- Improved detection consistency by 40% during internal evaluation through iterative
model tuning.
Technologies Used
Python
Google Colab
TensorFlow
Kaggle
Little Big Backrooms
Unity, C#, Blender
Project Overview
A first-person horror exploration game set in an endless maze of surreal,
liminal spaces inspired by the Backrooms creepypasta.
Key Contributions
- Overhauled the lighting pipeline and refactored inefficient game logic, boosting
performance from < 15 FPS to a stable 90+ FPS across all scenarios.
- Implemented seamless asynchronous level loading and resolved critical system bugs,
successfully transitioning the project from a broken build to a release-ready state.
- Expanded gameplay mechanics by programming dynamic horror effects and scripted
events, significantly enhancing player immersion and visual fidelity.
Technologies Used
Unity
C#
Blender
Zero-G Fitness (NASA Space Apps Lahore Challenge Winner 2024)
Unity, C#, Blender, Python
Project Overview
Developed a zero-gravity fitness application that utilizes body tracking
and pose estimation to provide real-time feedback on user exercises in a
simulated environment.
Key Contributions
- Engineered a cross-platform fitness application using Unity for platform
compatibility alongside OpenCV for robust body tracking and pose estimation.
- Implemented optimized C# Scripts to streamline data transfer and visualization
between the two interfaces, reducing required bug fixes by over 30%.
- Developed a dynamic avatar system from scratch, successfully mapping real-time user
pose data to 3D character rigs for instant visual feedback.
Technologies Used
Unity
C#
Blender
Python