Milad Mobaseri

Automation

AutoWin: Registration Automation for Competitive Platforms

A multi-part platform that automates registration on systems whose capacity fills within seconds, spanning a Windows launcher, an Android companion app, a central server and execution agents.

Period
Feb 2026 to present, in active development
My role
Sole designer and developer: architecture, server, client apps and agents

01The problem

Some registration systems open a limited capacity at a fixed moment, and many people compete for it at once. It fills within seconds, and anyone filling the form by hand usually loses their place before reaching payment.

These systems are not simple either: anti-bot detection, identity verification steps, crowded queues and unstable server responses. AutoWin automates the whole path from sign-in to the payment gateway, so the user enters their details once and completes the final payment themselves.

02Architecture

Central serverevery part connects here
  • Windows launcher⇄ server
  • Execution agentsstarted by the launcher
  • Android app⇒ server
  • Browser extension⇄ server

03Inside the product

  1. Windows launcher

    Where the user starts. It fetches bots from the server, stores accounts and per-system settings, and runs several jobs in parallel. Every run is visible as a live log. Each bot's guide is served from the server, so correcting it never requires a new release.

    Windows launcher
  2. Android companion app

    Relays the identity-verification step from the user's phone to the system, so a run never waits on manual input.

    Android companion app
  3. Browser extension

    For systems where part of the flow must run in the user's own browser; it works with the same account and the same server.

    Browser extension
  4. Server and agents

    The central server manages accounts, devices, versions and bot configuration; keys for external services are fetched only at run time, never shipped inside the app. Each agent is an independent run that reports diagnostic logs back to the server.

  5. Challenges

    Competing within fractions of a second means every extra request costs; bots are tested against a simulated environment before the real site. Anti-bot detection is handled by a local, purpose-built model with no external service. The core is protected, and internal errors are never shown to the user.

04In action

  • Real run; sensitive details blurred

  • Real run; sensitive details blurred

  • Real run; sensitive details blurred

05Stack

PythonDjangoC#.NETAvaloniaDjango REST FrameworkKotlinJavaScriptNginxGunicornCloudflare

06Recent changes

  1. Local purpose-built model for anti-bot detection, no external service
  2. Unified logging standard with diagnostic logs sent to the server
  3. Bots tested on a simulated environment before the live site
  4. Moved to a dedicated server and new backend without breaking installed versions
  5. External-service keys served only at run time

07What's next

AutoWin started with one kind of system and is expanding to other competitive platforms such as event ticketing. In progress: a monitoring panel with live logs and suspicious-activity detection.