Desktop AI Assistant (Electron + OpenAI API) — Real-Time Audio Transcription & Answering

Overview I'm looking for an experienced Electron desktop developer to build (and polish) a cross-platform AI assistant app. The app listens to live audio, transcribes it in real time using OpenAI Whisper, and streams back concise AI-generated answers using the OpenAI Chat API — all driven by the user's own API key. A working prototype exists; I need someone to harden it, add features, and package it into clean installers. What the app does - Captures system/loopback audio and microphone input in short windows. - Transcribes each window with OpenAI Whisper, with voice-activity detection to filter silence and reduce hallucinated text. - Sends the transcript (plus user-provided context) to the OpenAI Chat Completions API and streams the response into the UI. - Lets the user paste a screenshot and get an answer via the vision-capable models. - Includes an audio clean-up chain (high-pass/low-pass, compressor, noise gate) to improve transcription quality. - Frameless, always-on-top overlay window with a system tray icon and global hotkeys. Tech stack - Electron (main + renderer, plain HTML/CSS/JS — no framework/bundler currently) - Node.js, OpenAI REST API (chat completions with SSE streaming + audio transcriptions) - Web Audio API / MediaRecorder for capture - electron-builder for Windows (NSIS) and macOS (DMG) packaging - `pdf-parse` for resume/context PDF import Scope of work / deliverables 1. Review and refactor the existing prototype for stability and clean code. 2. Improve transcription accuracy (windowing, sentence assembly, noise handling). 3. Polish the UI/UX (settings, status indicators, error handling). 4. Ensure macOS system-audio capture works (guidance/helper setup). 5. Produce signed/packaged installers for Windows and macOS. 6. Short written setup + build documentation. Required skills - Proven Electron experience (main/renderer, IPC, preload, contextIsolation) - Strong JavaScript/Node.js - Experience integrating OpenAI (or similar) streaming APIs - Web Audio API / media capture experience - App packaging & code signing (electron-builder) Please start proposal with banana word to avoid auto bid.

ID del proyecto: 40569009