Cognify Architecture

🎮 Cognify 🎮

Cognify is an AI-powered word-guessing game where players compete against a sophisticated AI to guess a daily target word. Set in a futuristic world where AI has taken control of human knowledge, the stakes are high: players must outsmart the AI to reclaim and protect the words that define humanity's culture and history. Cognify challenges players to engage in a battle of wits and creativity, blending fun, education, and strategy. Together, human ingenuity and linguistic prowess become the last hope to save the world from an AI-driven erasure of identity.

Alt text

💡 Features

Alt text

1. AI-Driven Word Guessing

Compete against an AI that guesses alongside you using Amazon Bedrock Models.
Receive similarity feedback based on GloVe embeddings for each guess.

2. Interactive Gameplay

Alt text

Real-time updates via a Flask backend and JavaScript frontend.
A user-friendly interface with visual and audio feedback.

3. Daily Challenges

Alt text

A unique target word every day, challenging players to improve over time.
Leaderboards for competitive ranking.

4. Performance Tracking

Alt text

In-depth player statistics including total games played, guesses per game, and average completion time.
Save and review game history to track improvement.

5. Seamless Cloud Integration

Alt text

Hosted on AWS ECS and managed with CloudFormation, ensuring a scalable and secure experience.

🛠️ Architecture Overview

📂 ### Cognify Game Architecture

Architecture Overview

Frontend

User Interaction: Web Game UI built with HTML, JavaScript and CSS.
Key Features:
- Handles word input, feedback display, and leaderboard updates.
- Communicates with the backend API using AJAX.

Backend

Flask API

Processes:
- Player inputs.
- Game logic and AI integration.
Handles:
- Semantic similarity calculations.
- Ranking player guesses.
Interfaces with DynamoDB for data storage.

AI Integration

Amazon Bedrock:
- Mistral.mistral-7b-instruct-v0:2: AI-powered guessing logic.
- Titan Embeddings V2: Semantic similarity calculations.
Custom Logic: Ranks and optimizes similarity scores.

Dockerized Backend

Key Components:
- Flask API and AI logic encapsulated into Docker containers.
- Ensures portability and ease of deployment.

Hosting and Compute

Amazon ECR: Stores Docker images for frontend and backend.
Amazon ECS: Orchestrates containers in production.
Amazon EC2: Provides compute instances for running containers.

Amazon Q (CodeWhisperer)

Capabilities:
- Assisted in coding efficient Flask API and Dockerfiles.
- Optimized container configurations and AWS service integration.
- Suggested scalable deployment solutions.

Database

Amazon DynamoDB:
- Stores player profiles, game history, and statistics.
- Ensures low-latency data retrieval for smooth gameplay.
- Manages leaderboard and daily challenges.

AI Processing Workflow

Player Input: Sent to backend for processing.
AI Guessing:
- Titan Embeddings V2: Embedding similarity calculations.
- Mistral.mistral-7b-instruct-v0:2: AI-generated guesses.
Ranked Results: Returned to players in real-time.

Game Hosting and Infrastructure

Docker Containers:
- Frontend and backend services packaged for simplicity.
- Streamlined deployment and scaling.
Elastic Container Service (ECS): Auto-scales to handle traffic.
Elastic Load Balancer: Distributes requests efficiently.
Amazon EC2: Runs the game server and backend logic.
Route 53: Manages game domain DNS.

Player Analytics and Statistics

Tracks:
- Total games played.
- Best rank achieved.
- Completion rate and recent game activity.
Visualized on the user dashboard for insights.

Development Automation (Amazon Q)

Key Contributions:

Generated Flask API routes, logic, and frontend components.
Facilitated integration with AWS services (Bedrock, ECS, DynamoDB).

Optimization Areas:

Semantic similarity calculations.
AI inference pipeline.
Dockerfile creation and deployment automation.

Security and Scalability

Security Measures:
- Data encryption for storage and communication.
- Role-based access control for sensitive resources.
- Secure API Gateway interactions.
Scalability:
- Auto-scaling policies ensure high availability.

👥 Team

Cognify was developed by a passionate team of developers and AI enthusiasts committed to creating a fun, educational, and competitive game experience.

🛠️ Challenges We Ran Into

Alt text

Embedding Model Limitations: GloVe embeddings' limited vocabulary required additional fallback logic.
Optimizing AI Performance: Reducing latency for real-time gameplay involved extensive caching and optimization.
Infrastructure Setup: Deploying and scaling on AWS ECS and integrating Route 53 presented initial challenges.

🏆 Accomplishments That We're Proud Of

Alt text

Successfully integrating GloVe embeddings and Amazon Bedrock Models for dynamic word guessing.
Creating a visually appealing and responsive interface for an engaging user experience.
Deploying a scalable cloud-based solution that ensures reliability and performance.

🎓 What We Learned

Alt text

Natural Language Processing: Gaining insights into embedding techniques and similarity calculations.
Cloud Infrastructure: Building and managing scalable solutions using AWS services.
User-Centric Design: Iterating on feedback to refine the gameplay experience.

🌟 What's Next for Cognify

Multiplayer Mode: Enable real-time competition between friends.
Mobile App: Launching a mobile version for iOS and Android.
Enhanced AI: Using more sophisticated NLP models to improve the AI's guessing ability.
Gamification Features: Adding achievements, rewards, and daily streaks to boost player engagement.
Language Support: Expanding vocabulary and support for multiple languages to reach a broader audience.