Mehtab Mahir
Mehtab Mahir
Aspiring Software Engineer
Queens College, NY
I'm Mehtab Mahir, a Queens College Computer Science graduate (Dec 2025) with a passion for turning curiosity into polished, practical software. Growing up, I built PCs, worked on my car, and even my guitars to understand how things work and make them better. That same instinct now guides how I design tools that are both approachable and high-performing.
I specialize in cross-platform desktop apps and AI-powered features that stay friendly to beginners. Projects like EasyWhisperUI sharpened my focus on performance, user onboarding, and thoughtful automation.
Whether I'm collaborating in a hackathon or leading a team project, I prioritize clear communication, pragmatic architecture, and building software people genuinely enjoy using.
Projects
EasyWhisperUI
Cross-platform desktop app that makes local Whisper (whisper.cpp) transcription beginner-friendly with GPU acceleration on Windows and macOS. Migrated the UI to Electron (React + TypeScript) with a secure IPC bridge for privileged operations. Automated Windows setup installs/validates dependencies and stages whisper.cpp for Vulkan acceleration (Intel/AMD/NVIDIA). Supports queue-based batch transcription and live transcription (whisper-stream) with streamed output/events back to the UI. Integrates FFmpeg for media preprocessing and automatic Whisper model downloads.
Auto 60hz C++
Lightweight C++ app that automatically switches between 60Hz and 120Hz (or custom) refresh rates based on active applications. Built using WinAPI and PSAPI for fullscreen detection, with a modern Qt interface, custom refresh values, and startup integration. Runs silently with zero CPU usage, designed for efficiency and battery savings.
Telemedicine App (YHack 2024)
Built in 24 hours with three others at YHack 2024. AI-powered chatbot for secure medical consultations using OpenAI, Firebase, and Next.js. Real-time chat between patients and healthcare providers, wrapped in a responsive Material UI and backed by a solid, scalable backend.
Database Normalization Project
Led a team of 6 for a SQL Server project that turned raw sales data into a fully normalized star schema. Included ERDs, automated load scripts, sequence-based keys, and stored procedures. Focused on clean schema design, integrity constraints, and real-world business logic.
Gemini Chatbot
React-based chatbot leveraging the Google Gemini AI API, integrating advanced image, audio, and Google Cloud Speech-to-Text API input capabilities.
Machine Learning with Python
A PyTorch-based reinforcement learning project where an AI agent learns to play Snake by interacting with the game environment. The agent uses a neural network (Linear_QNet) to predict moves, stores experiences in replay memory, and balances exploration with exploitation as it trains. After each game, the agent restarts automatically and continuously improves its score by learning from both short- and long-term memory. Training progress and scores are visualized, making the learning process transparent.
Skills
Desktop & Web Application Development
Custom cross-platform desktop and web apps with a focus on performance, usability, and reliability.
AI & API Integration
Integration of advanced APIs and machine learning models (OpenAI, Google, Spotify, Reddit, more) into modern apps.
UI/UX & Automation
Modern UI/UX design, process automation, and installer creation for effortless deployment.