● Built a full-stack web application for real-time speaker diarization using Next.js, FastAPI, and Whisper ASR, with VAD and speaker embedding clustering for live transcription and speaker identification
● Developed offline refinement pipeline for high-accuracy diarization, generating downloadable transcripts in PDF, JSON, and SRT formats with customizable speaker labels
● Implemented a responsive frontend with WebSocket audio streaming, live captions, and export features, supported by Redis for session state management and S3 storage for artifacts
● Containerized all services with Docker and orchestrated deployment on AWS ECS/EKS for scalable, production-ready infrastructure
Next.js | FastAPI | Whisper | Redis | Docker | AWS
Speaker Diarization App
● Authored Python Simplified with Generative AI, a beginner-friendly guide to programming with Python using Jupyter Notebooks, published in both print and digital formats.
● Collaborated with two co-authors and an editorial team to ensure clarity, technical accuracy, and pedagogical structure for a self-paced learning experience
● Leveraged AI-enhanced workflows to demonstrate modern development practices, including prompt engineering, AI-assisted test writing, and project-based learning.
BPB Publications
Python Simplified with Generative AI
● Developed a full-stack voice recognition application using Python FastAPI and React/TypeScript, implementing real-time speech-to-text with faster-whisper ASR
● Engineered voice command processing with push-to-talk interface supporting commands like "white 3 goal at 2:34, assist 9" with deterministic regex parsing and real-time event confirmation
● Built NCAA-compliant scoresheet system with PyMuPDF PDF generation, OCR roster extraction, and coordinate-based data stamping for official tournament documentation
● Implemented comprehensive state management with real-time scoreboard, period tracking, event history, and undo functionality using Zustand and React Query
Python | FastAPI | React | TypeScript
Voice-Driven NCAA Score Sheet
● Designed a full-stack file sharing application using Next.js with TypeScript, implementing
secure file storage with Convex backend and real-time data synchronization
● Engineered user authentication and organization management using Clerk, with role-based access control (admin/member) and secure file permissions
● Developed responsive UI with Tailwind CSS and shadcn/ui components, featuring file previews, favorites system, and advanced search/filtering capabilities
● Implemented file upload system supporting multiple formats (PDF, CSV, images) with automatic type detection and secure storage using Convex file storage
Next.js | Convex | Clerk Authentication
Note-Share
● Built a full-stack financial management platform using Go and Node.js with SQLite database, implementing secure authentication with JWT tokens and bcrypt password hashing
● Engineered responsive web interface with interactive dashboard, real-time transaction tracking, and campus dining menu with dietary restriction filtering (vegetarian, vegan, gluten-free)
● Developed normalized database schema with foreign key relationships for users, transactions, and budget settings, ensuring data consistency and performance
● Implemented budget tracking system with configurable spending limits, transaction categorization, and automated notifications for better financial awareness
HTML | JavaSCript | Node.js | SQLite | JWT
FlexiBudget
● Designed an course discovery platform that transforms course catalogs into conversational search experiences
● Implemented semantic search using ChromaDB vector database with specialized collections, enabling fuzzy matching across 2,700+ course records and natural language queries like "What CS courses can I take with Professor Smith?"
● Integrated OpenAI GPT-4 API for contextual responses with schedule conflict detection, reducing student registration time by 70%
● Built automated data pipeline processing CSV catalogs into searchable vector embeddings and developed multi-modal query routing for comprehensive academic planning scenarios.
Go | OpenAI GPT-4 API | ChromaDB | Docker | Vector DB
AI-Powered Course Search
Let's connect: @evan_haba