Muhammad
Hasnain

Classic arcade brick-breaker with 50 hand-crafted levels and progressive difficulty scaling.

Problem

Design a complete game loop with consistent difficulty scaling and genuine replayability across 50 distinct stages.

Approach

Used Unity's physics engine with custom ball behavior, multi-ball mechanics, and responsive paddle input — tuned to feel tight, not floaty.

Systems

Level data architecture via ScriptableObjects, particle feedback system, progressive speed/ball-count scaling controller, and score persistence.

Outcome

Deployed as a playable WebGL title on EFG Games platform.

Time-pressure puzzle game requiring precise timing judgments across 20+ handcrafted levels.

Problem

Build tight, satisfying timing mechanics that feel responsive and fair — failure should feel earned, not random.

Approach

Built a tick-accurate timer system with layered visual cues so players always know where the beat is, even under pressure.

Systems

Custom clock animation rig, adaptive difficulty curve controller, streak-based score multiplier, and session persistence.

Outcome

Live on EFG Games platform with regular player traffic.

Phonics-based educational game designed for early literacy on the NeuroNovels platform.

Problem

Create a phonics game that adapts to learning pace without overwhelming or boring the player — classroom-tested requirements.

Approach

Designed adaptive feedback loops tied to correct/incorrect answer streaks; correct answers accelerate progression, errors provide targeted hints without breaking flow.

Systems

Audio-visual phonics matching engine, difficulty state machine, accessibility-aware UI with configurable text size, and session progress saving.

Outcome

Live on NeuroNovels educational platform serving real learners.

Applied ML pipeline for real-time surface defect detection on manufactured parts. NDA client project.

Problem

Client needed automated, reliable defect detection on manufactured surfaces with a 3D visualization layer for operator review.

Approach

Designed and trained a custom ML classification model on client-supplied defect imagery; paired inference output with a 3D visualization renderer for operator-facing review.

Systems

Full ML pipeline (data preprocessing → training → inference API), 3D surface visualization layer, and integration interface for existing client tooling.

Outcome

Deployed to client production environment. No public repo or demo — NDA applies.