AWS Builder Center

Weekend Showcase Challenge: Smart Downloads

Smart Downloads was built to solve exactly that: an app that runs in the background, watches the Downloads folder in real time, and as soon as a new file finishes downloading, it automatically.

Weekend Showcase Challenge: Smart Downloads

Vision & What It Does

Every week I lost time to the same annoying task: downloading a file, forgetting about it in my Downloads folder, and weeks later ending up with hundreds of PDFs, images, installers, and videos all mixed together with no organization at all. Sometimes I couldn't even remember if I had backed up something important — a resume, a contract, a receipt — and that turned into a small but recurring source of anxiety.
Smart Downloads was built to solve exactly that: an app that runs in the background, watches the Downloads folder in real time, and as soon as a new file finishes downloading, it automatically:
  1. Identifies the file's category (Document, Image, Video, Audio, Archive, Installer)
  1. Moves the file into a category-organized subfolder
  1. Backs it up to Amazon S3
From the user's perspective, the experience is simple: open the app once, configure your S3 bucket and which categories to organize, and from then on just download files normally. A live dashboard shows the monitoring status, the last file processed, how many files were organized today (broken down by category), and a history of recent events — so you always know backups are happening, without having to think about it.




How I Built It

I split the project into clearly separated modules, to avoid one giant, hard-to-maintain file:
  •  config.py  — settings and categories
  •  monitor.py  — folder watching
  •  organizer.py  — categorization and file-moving logic
  •  uploader.py  — S3 integration
  •  ui.py  — graphical interface
  •  utils.py  — logging and history
The most important architecture decision was using the  watchdog  library to monitor the file system in real time, instead of polling the folder at intervals, which would be slower and waste resources.
The biggest technical challenge showed up after my first round of testing: files organized manually worked perfectly, but files actually downloaded from the internet weren't being organized at all. After investigating, I found that browsers like Chrome first save downloads as a temporary file ( file.pdf.crdownload ) and only rename it to the final filename once the download completes. My code was only listening for file creation events, so it was ignoring that rename entirely.
The fix: also handle watchdog's  on_moved  event, and wait for the file size to stabilize before moving it, to avoid grabbing a half-finished download.
Another detail worth mentioning: avoiding overwrites. If a file named  photo.png  already exists in the destination folder, the app automatically saves the new one as  photo (1).png .

Rebuilding the Infrastructure as Code

For the showcase, I took the infrastructure side further and rebuilt the AWS setup as Infrastructure as Code with Terraform, instead of clicking through the console. A single  terraform apply  now provisions:
  • A private, encrypted, versioned S3 bucket with public access fully blocked
  • A dedicated IAM user with a least-privilege policy scoped only to that bucket (list/get/put — nothing else, not even bucket creation)
This makes the whole environment reproducible from scratch in minutes, and it's a much safer pattern than a long-lived admin-level key sitting in a config file.
One small but real lesson from redeploying this way: after applying the Terraform, the app's settings still pointed to an older bucket name from a previous manual test. Nothing failed loudly — it just silently wouldn't have found the right destination. Double-checking that the bucket name in the app's settings matches the  bucket_name  Terraform output exactly is now a step I won't skip again.

What's Next: The Backup Center

The current dashboard already surfaces monitoring status, the last file processed, and today's history — the natural next step is turning that into a dedicated Backup Center view:
  • ☁️ Total files backed up
  • 💾 Total backup size
  • ✅ Last successful backup
  • ❌ Failed uploads
  • 🟢 Overall status: "Cloud Backup: Healthy"
It turns the pitch from "I organize your downloads" into "I organize your downloads and show you they're protected in the cloud." That's the next iteration I'm planning to build on top of this showcase version.

AWS Services & Architecture

  • Amazon S3 — stores the backup of organized files, with prefixes per category (e.g.  bucket/Documents/file.pdf )
  • IAM — dedicated user with permissions scoped to S3 operations on a single bucket, used by the app's local credentials via  boto3 

 smart-downloads-backup-gui-ago2026  bucket in the AWS console — category folders (Archives, Audio, Documents, Images, Installers, Videos), and inside  Images/ , files uploaded automatically by the app
Architecture flow:
Downloads folder
→ watchdog detects a new file (created/moved event)
→ organizer.py identifies the category and moves it to the local organized folder
→ uploader.py uploads the file to Amazon S3 (bucket/<category>/file)
→ utils.py logs the event to history
→ ui.py updates the live dashboard

What I Learned This Summer

The main technical lesson was about how browsers actually handle file downloads on disk — something I hadn't accounted for in the initial design, and that only surfaced when testing with real downloads instead of simulated file creation. It reinforced how important it is to test the complete end-to-end flow before considering a feature done.
On the AWS side, this was my first time integrating  boto3  in a practical way inside a desktop app with a graphical interface, including credential error handling — a solid first hands-on experience with how a desktop application can connect to cloud services without needing a separate backend.
Rebuilding the environment with Terraform for this showcase pushed that further: moving from console clicks to a reproducible, least-privilege infrastructure definition is a habit I'm carrying into every project from now on.

Try It Yourself

Builder Who Inspired Me

Shoutout to David Vergara ( @davidvl ) for Daily Music Drop, his entry for the Creative Agent weekend. What stuck with me was how small the architecture was for something genuinely "always-on" — EventBridge Scheduler triggering a Lambda that calls Bedrock once a day, no server, no manual prompt, just a fresh recommendation waiting on the site every morning. It's a good reminder that an agent doesn't need to be complicated to be interesting, and it nudged me to think about how Smart Downloads could work the same way: running quietly in the background and just being there when you need it, without asking anything of the user.
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article