andrew khadder
Memory Sharing MCP Server
Python, FastMCP, Redis, Next.js
  • Awarded 3rd place in the MCP & Agents Hackathon, recognized for innovative cross-agent memory sharing
  • Architected and implemented an MCP server, enabling seamless context sharing across multiple AI clients
  • Integrated Redis vector store for storage; supporting both short-term and long-term knowledge retention
  • Built web dashboard for managing API keys and memory records; enhancing user control and observability
AI Playlist Art Generator
Python, FastAPI, SciKit-learn, Dall-E, ChatGPT
  • Developed a backend app that analyzes Spotify playlists and generates custom AI art based on its mood
  • Built and trained custom machine learning classification model using track metadata and genre embedding
  • Engineered modular ML pipeline for preprocessing, feature extraction, training, evaluation, deployment
  • Integrated Spotify Web API for user authentication and dynamically retrieved playlist metadata to power analysis
AI Leetcode Tutor
React, Chrome API, Gemini AI API
  • A Google Chrome Extension that gives AI-powered contextual hints directly onto Leetcode problems
  • Implemented content scripts to scrape the problem statement and code to query to deliver personalized guidance
  • Integrated Google Gemini via secure background workers, while keeping all user code on-device for privacy
AI Song Recommender
Next.js, Tailwind CSS, OpenAI, Spotify API
  • A website that helps streamline hackathon management by providing a dashboard for users and admins
  • Optimized backend API routes with serverless functions to reliably scale and support 500+ concurrent users
  • Currently in use by 5+ hackathon organizations on campus to host, maintain, and promote their events
  • Implemented virtualized infinite scrolling on all dashboard tables reducing server load and queries by 30%