Social Peak
Desktop tool that automates tasks across several social media accounts at once, with an independent module for each platform.
- Client
- Personal Project
- Year
- 2025 – 2026
- Service
- AI and automation
- Role
- Full-stack technical lead
- Format
- Desktop app
- Status
- Internal

Automating several social media accounts at once without mixing their sessions.
We needed a desktop tool that ran tasks in parallel on several Instagram, Facebook, X, TikTok, YouTube and Kick accounts, with every session isolated and progress visible in real time.
A modular tool with a desktop launcher.
We built the tool in Python on top of SeleniumBase and Flask, with an independent module for each platform that is updated separately. Every account runs with its own isolated profile so sessions never cross.
We packaged it for Windows with PyInstaller and a graphical launcher with a loading screen, a system tray icon and automatic start of the Flask server. Every task sends its log in real time to Universe, the central panel.
What it includes
Several accounts in parallel
Threads and multiprocessing in Python to run several accounts at once without mixing their data.
One module per platform
Independent Flask blueprints for Instagram, Facebook, X, TikTok, YouTube and Kick, updated separately.
Desktop launcher
Animated loading screen, system tray icon and automatic start of the Flask server.
Real-time log
Every task sends its progress over WebSocket to the Universe central panel.
Spanish and English
Interface translated into both languages with Flask-Babel.
Built-in updater
Visual updater included in the tool to receive every new version.
Stack and architecture
- Python 3.11
- SeleniumBase
- Flask
- Flask-Babel
- Supabase (PostgreSQL)
- Socket.IO
- customtkinter
- PyInstaller
- Desktop
- Windows executable built with PyInstaller and a launcher made with customtkinter and pystray.
- Backend
- Python 3.11 with SeleniumBase and Flask, with one blueprint per platform.
- Real time
- Socket.IO client that sends every task's log to the Universe server.
What we solved
Parallel sessions without crossovers
Centralized state with a Singleton pattern and one profile per account in separate temporary folders, so logins never mix.
Memory and performance on Windows
We release idle drivers to save memory and package everything with PyInstaller into an optimized executable.
In detail
- Platforms with their own module
- 6
- Executable size
- ~190 MB


