Integrated ML detection outputs into Unity with a clean separation between inference results and runtime visualization: config-driven marker placement and event-driven updates for maintainability.

Built and shipped a suite of production WebGL learning games using a reusable systems-first architecture (state/progression/save-load) and a content pipeline that reduces per-game integration time.

ScriptableObject-driven level configs and difficulty curves, robust collision handling, scoring/combo logic, and clean game-loop architecture designed for content scaling.

A polished 2048-style puzzle with smooth input, grid animation, score tracking, restart/undo-ready state design, and WebGL performance tuning for consistent frame pacing.

Word-search grid game with dynamic board generation, selection path logic, word validation, and a scalable content setup for adding new word packs without code changes.

Sudoku with validation rules, hint/error feedback, clean UI flow, and puzzle-generation/selection support—implemented with modular board logic and reusable UI components.

Interactive analog clock with drag/rotate input, time validation, feedback loops, and ScriptableObject-driven question sets—built for reuse across multiple learning modules.

Math quiz game with scramble-style problems, timer, score tracking, and configurable difficulty—built for WebGL with clean state and feedback systems.