Verbal Aptitude Qns
General Assistant
Tech Coach (DSA)
DSA qns
Logical Reasoning qns
CBT Exam Simulator
Quantitive Aptitude qns

Inspiration

What it does

🌟 Inspiration

Traditional voice assistants are stateless command-receivers—great for timers and music, but completely ineffective for serious technical interview preparation. When a developer asks traditional Alexa to explain code, it recites curly braces aloud or forgets the candidate's past struggles five seconds later.

With Amazon opening up Alexa+ and the Model Context Protocol (MCP), we realized we could transform Alexa from a simple voice assistant into an interactive, multimodal technical mentor designed for the Amazon Echo Show.

Our vision: An AI coach that listens to your verbal reasoning, evaluates Big-O complexity, renders syntax-highlighted IDE code on screen while keeping voice answers crisp, and tracks your weaknesses across sessions using Amazon DynamoDB.

🚀 What it does

Alexa+ Tech Coach acts as a live FAANG-style whiteboard coding and aptitude interview partner:

🗣️ Multimodal Separation: Alexa speaks concise, conversational guidance (2–3 sentences) while visually rendering a split-screen LeetCode IDE or CBT Exam Arena on the Echo Show screen.
🧠 Verbal Logic & Complexity Evaluation: Evaluates spoken algorithm reasoning (e.g., "I'll use a two-pointer approach"). It analyzes $O(N)$ time & $O(1)$ space complexity in real-time before writing code.
📦 Long-Term Memory via Amazon DynamoDB: Remembers weak topics, streak counts, and past interview mistakes across sessions so every daily problem targets candidate weak spots.
📝 Amazon OA (CBT) Exam Simulator: Simulates Amazon Online Assessments with timed Quantitative, Logical, and Verbal aptitude test cards featuring hints, negative marking, and derivations.
🛑 Voice Barge-In: Real-time conversational interruption allowing users to cut off Alexa instantly to give their answer without waiting for speech playback to finish.

💡 Traditional Alexa vs. Alexa+ Tech Coach

Feature	Traditional Alexa	Alexa+ Tech Coach
Brain / Reasoning	Rigid Slot-Intent rules	Generative AI Reasoning (AWS Bedrock / Groq)
Code Presentation	Speaks raw syntax aloud	Multimodal Separation: Speaks concepts, renders IDE code
Personalization	Stateless (forgets every session)	Amazon DynamoDB Memory: Tracks streaks & weak topics
Interview Coaching	Cannot evaluate verbal logic	Verbal Complexity Analysis: Evaluates $O(N)$ vs $O(N^2)$
Tool Extensibility	Closed ASK Skills	Model Context Protocol (MCP) + Real-time Web Search

🏗️ How we built it

Self-Hosted MCP Server (Python & FastAPI): Implemented an MCP server conforming to the official Model Context Protocol specifications (SSE transport) exposing tools:
- get_daily_problem: Dynamically fetches algorithm/aptitude challenges based on candidate history.
- evaluate_logic: Evaluates verbal logic and time/space complexity.
- fetch_tech_news: Pulls real-time tech updates.
AI Orchestrator Backend (Node.js & Express): Manages intent classification, crash-guard fallbacks, and real-time live web search integration.
Cloud Intelligence & Long-Term Memory:
- AWS Bedrock / Groq Llama 3 for generative reasoning.
- Amazon DynamoDB for persistent candidate profiles, streaks, and weakness tracking.
Multimodal Frontend (Next.js 16 & TypeScript): Tailored Echo Show UI with Web Speech API for voice recognition, speech synthesis, and dark-mode coding environments.

🧗 Challenges we ran into

Audio Overlap & Latency: In technical interviews, candidates often speak while the assistant talks. We implemented an instant Voice Barge-In listener that cuts synthesizer audio immediately when the user speaks or taps the microphone.
Multimodal Cleanliness: Ensuring Alexa doesn't read out raw syntax or bracketed code; separating voice output from visual display payloads required robust structured JSON output schemas.
Self-Hosted MCP Integration: Bridging the Python MCP SSE transport with our Next.js/Node.js client ecosystem required custom protocol handling.

🏆 Accomplishments that we're proud of

Full working implementation of a self-hosted Model Context Protocol (MCP) server for Alexa+.
Seamless multimodal experience: crisp spoken voice synced with real-time code editor and CBT exam cards.
Adaptive DynamoDB weakness engine that genuinely personalizes daily challenges.

📚 What we learned

How MCP standardizes tool calling and external knowledge retrieval for the next generation of conversational AI.
The power of multimodal separation—voice is for conversation, screens are for visual comprehension.

🔮 What's next for Alexa+ Tech Coach

Direct camera integration for real-time paper-and-pen whiteboard OCR scanning.
Expanded company-specific mock interview tracks (Amazon Leadership Principles Q&A, System Design diagrams).