Savin De Alwis

Building a minimal, local-first desktop focus app designed to enhance deep work sessions through intelligent activity tracking and distraction control.

• Engineered active window tracking with real-time classification into Deep Work, Working, or Distraction tiers to dynamically adapt focus-session behavior and enforce optional distraction blocking.
• Architected a lightweight, secure desktop client using Tauri, Rust, and SQLite for private, local-first data storage and high-performance execution.
• Built productivity tooling including ambient audio generators, lightweight desktop overlays, session history, and dashboard analytics with React and TypeScript.

Built a full-stack platform for Model United Nations operations, covering enrollment, marking, and performance analysis for live conferences.

• Used in COMUN 2025 and 2026 to track 500+ delegates
• Designed a fairness-focused judging algorithm based on prior recognition and performance signals.
• Built a hybrid workflow with a CLI for rapid data entry and a Django REST API backed by Firestore.

Developed a production-ready feedback platform for youth conferences, with a strong emphasis on UX quality and role-sensitive workflows.

• Led UX design and database architecture while collaborating with engineering and marketing stakeholders.
• Built the app in Next.js with a focus on responsive multi-page flows and polished admin experiences.
• Implemented attribute-based access control to support granular event-level permissions.

Building a WhatsApp-native AI assistant that turns everyday conversations and forwarded content into useful, structured actions.

• Unifies calendar, reminders, to-dos, finance tracking, and email management in the channel users already use every day.
• Proactively extracts structured data from forwarded content, eliminating repetitive manual input.
• Developing a cross-platform experience with Next.js, React Native, and AWS.

Building an interactive disaster-response simulator for exploring urban emergency scenarios, resource movement, and planning tradeoffs in real time.

• Simulates accidents, floods, epidemics, and blocked-road events across a synthetic city network.
• Combines an interactive map and dashboard to track resource allocation, bottlenecks, and emergency-unit load.
• Exploring heuristic-based response planning to evaluate strategy changes under different constraints.

Built an interactive visualizer to deepen my understanding of pathfinding and maze-generation algorithms.

• Implemented BFS, DFS, Dijkstra's, and A* from scratch using queues, stacks, priority queues, and graphs.
• Added weighted terrain, maze generation, and runtime comparisons to make algorithm tradeoffs visible.
• Designed as a teaching aid for myself and other students revisiting these concepts.

Built a Wordle-solving algorithm around information theory to explore optimization and search strategy.

• Used entropy and word-frequency heuristics to produce stronger guesses than the average human player.
• Measured results at roughly 12% better than typical human performance.
• Expanded the project into my Edexcel Extended Project Qualification.

Designed and built this portfolio as part of the Cloud Resume Challenge, combining product presentation with cloud deployment.

• Built with Next.js and Tailwind CSS for a fast, responsive frontend.
• Deployed on AWS using S3, CloudFront, and Route 53 for reliability and performance.
• Used this project to sharpen visual design, frontend polish, and infrastructure delivery.

Built a 2D adventure game as a hands-on introduction to core game development systems.

• Implemented collision systems, map generation, and enemy behavior in Python with Pygame.
• Added inventory-style mechanics such as chests, sprinting, and weapon pickup/drop interactions.
• Used the project to learn how multiple gameplay systems interact in a real codebase.