E-Mart
Android e-commerce app with an animated splash screen, a daily-offers dashboard, and SMS alerts on purchase.
JavaView on GitHub →
M = (Q, Σ, δ, q0, F)
Ahmad Mubeen
BSCS student and application-development trainee — currently teaching a computer to tell real motion from generated motion, one cloth simulation at a time.
q1 — about
From game worlds to ground truth
I'm studying Computer Science (BSCS) at FAST‑NUCES, Lahore. I actually started in game development — taught myself Unreal Engine 5 and shipped a few working prototypes on my own — but there wasn't much of a game-dev community or mentorship around at FAST, so I redirected that energy toward AI and software development, which is where most of my work lives now.
Right now I'm a Trainee in Application Development at Stewart Pakistan, and running a YouTube channel called Debugging Myself, where I turn my own coursework into playlists for anyone stuck on the same problem set. Outside of code: badminton, chess, and — most recently — German, which I'm learning through Duolingo and Udemy with an eye on future opportunities abroad.
q2 — final year project · ongoing, target 2027
AI Fake Video Detection Through Cloth Simulation
Most deepfake detectors chase pixel-level artifacts left behind by a specific generator — which means they age badly as generators improve. This project takes a different angle: it checks whether a person's clothing moves the way real fabric does. Gravity, stretch, drape, and collision with the body follow physics that generative video models don't explicitly simulate, so the gap between simulated and observed cloth motion should generalize across generators, rather than overfitting to one model's cosmetic fingerprints.
- doneFrame extraction from raw video
- doneCloth segmentation (SAM2) & body pose extraction (MediaPipe)
- nextCloth point tracking across frames
- plannedPhysics-based cloth simulation & residual computation
- plannedClassifier training & evaluation
- plannedDemo interface
q3 — projects
Things I've shipped, and one I'm still building
TrackEx
An expense-tracking app for Android, built on a Room database with a deliberately minimal UI.
JavaView on GitHub →
PuzzleYou
Capture a photo and it becomes a playable 4×4 sliding puzzle — image processing meets a simple game loop.
Kotlin / JavaView on GitHub →
Road Pulse
An Android app in active development — built from UI mockups up, with 29 commits and counting.
JavaView on GitHub →
Anonymous Messaging idea stage
A personal side project: send a message to anyone without revealing your identity, and without it ever sitting on a server's disk. Early days — building it in spare time.
In progress
q4 — experience & education
Where I've been putting the hours in
-
Present
Trainee, Application Development
Stewart Pakistan
-
In progress
BSCS, Computer Science
FAST‑NUCES, Lahore
q5 — teaching
Debugging Myself, on YouTube
Learning sticks better once you've had to explain it. That's the premise behind the channel: full course playlists built from my own semester material, published for anyone else stuck on the same problem set.
q6 — beyond code
Off the clock
Badminton & chess
Two very different games — one fast and reflexive, one slow and calculated. I like keeping both in rotation.
Game dev roots
Before the pivot, I self-taught Unreal Engine 5 and built a handful of prototypes — a few are up on my LinkedIn.
Learning German
Working through Duolingo and Udemy in parallel, with future opportunities abroad in mind.
q7 — contact · accepting state
Let's talk
Open to internships, collaborations, and conversations about deepfakes, Android, or why FAST could use a game-dev club.