A clean, fast, open-source alternative to Google Drive + Google Photos. Just create an account and start uploading.
Nimbus is a hosted, free-to-use cloud drive: a place to keep your documents, photos, and videos, browse them in a polished web app, and get them back on any device. It looks and feels like Drive and Photos — folders, a photo timeline, sharing, search — without the subscription and without a big company mining your library.
It's also fully open source. The whole thing runs on a serverless, pay-nothing-while-idle architecture, so anyone can read the code, learn from it, or contribute. But you don't need to know any of that to use it — just sign up and upload.
Your photo library — thumbnailed automatically, browsable in a fast grid, with uploads running in the background.
Files & folders
- Upload files or whole folders, download, preview, rename, move, and organize
- Soft Trash with restore — nothing is gone until you say so
- Star important items and give folders colors
- List and grid views, sortable, with infinite scroll
Photos & media
- A Photos timeline with a full-screen lightbox (zoom, slideshow, cast to TV)
- In-browser photo editing (crop, rotate, brightness, contrast)
- Automatic thumbnails for every image, generated server-side
- Duplicate detection and photo stacks; map view from photo GPS data
Uploads that don't give up
- Big files upload in resilient multipart chunks
- A Google-Drive-style upload tray with per-file pause / resume / cancel
- Drops on flaky networks auto-pause and resume when you're back online
- Resume interrupted uploads, and bulk-import a Google Takeout export
Sharing & collaboration
- Share links — public, expiring, or to named recipients — for files and folders
- A Shared with me view
- File versioning with bounded history
Find & track
- Full search with filters and saved searches; Recent view
- An Activity feed and login-activity log
- Storage usage breakdown by type
Your account, your way
- Email/password or Google / GitHub sign-in
- Session management — see and revoke signed-in devices
- Light / dark theme, English / हिन्दी, and accessibility-minded UI
Every feature listed above is built and running today — not a roadmap.
 My Cloud — files, folders, and storage at a glance |
 Activity — a timeline of everything you've done |
 Shared with me — folders others sent your way |
 Trash — soft-deleted, restorable until purged |
 Settings — theme, language, and preferences |
 Resume files - resumable uploadation without any issues |
- Create an account at nimbus-ochre.vercel.app — email/password, or Google / GitHub.
- Upload files, folders, or drag-and-drop right onto the page. Big files and slow connections are handled for you.
- Organize with folders, stars, and colors; browse photos in the timeline.
- Share a link when you need to — public, expiring, or to specific people.
- Find anything with search, Recent, and the Activity feed.
That's it. There's nothing to install and nothing to configure.
Nimbus is fully serverless and costs nothing while idle. The API is a single AWS Lambda (FastAPI behind Mangum) fronted by an API Gateway HTTP API. Because Lambda is stateless, the database (MongoDB Atlas) and file storage (S3) live outside it. Crucially, file bytes never flow through the API — the browser uploads and downloads straight to S3 using short-lived presigned URLs, so compute cost stays near zero no matter how large the files are.
flowchart TD
subgraph Client
B["Browser<br/>Next.js web app"]
end
subgraph AWS["AWS · ap-south-1"]
GW["API Gateway<br/>HTTP API"]
L["Lambda: nimbus-api<br/>FastAPI + Mangum"]
S3[("S3 bucket<br/>private · SSE-S3")]
T["Lambda: nimbus-thumbnailer<br/>Pillow"]
end
DB[("MongoDB Atlas<br/>metadata")]
B -- "JSON / auth (JWT)" --> GW --> L
L <-- "users, items, metadata" --> DB
L -- "issue presigned URLs" --> B
B -- "upload / download bytes (presigned)" --> S3
S3 -- "ObjectCreated event" --> T
T -- "512px JPEG → thumbnails/" --> S3
What happens when you upload a photo
- The browser authenticates and holds a short-lived JWT (with a refresh cookie so a reload keeps you signed in).
- It asks the API for a presigned
PUTURL and sends the bytes directly to S3 — the Lambda only records metadata. - An S3
ObjectCreatedevent triggers the thumbnailer Lambda, which writes a 512px JPEG underthumbnails/. - To view or download, the browser requests a presigned
GETURL and fetches the bytes straight from S3.
| Layer | Technology |
|---|---|
| Frontend | Next.js 16 (React 19), static export, client-rendered, Lucide icons |
| Backend | FastAPI (Python 3.13) + Mangum on AWS Lambda |
| API edge | AWS API Gateway (HTTP API) |
| Database | MongoDB Atlas via Motor (async) |
| File storage | AWS S3 — private bucket, SSE-S3, presigned direct upload/download |
| Thumbnails | S3 event → Lambda (Pillow) |
| Auth | JWT (HS256) + bcrypt, httpOnly refresh cookie, Google/GitHub OAuth |
| Infrastructure | CloudFormation (infra/nimbus-backend.yaml) |
| CI/CD | GitHub Actions — CI on every PR; deploy on merge via GitHub OIDC (no stored AWS keys) |
Base path: /api/v1. All /files routes require a Bearer JWT.
Auth — /auth: register, login, refresh, logout, me, OAuth
callbacks, session management, login-activity.
Files — /files: paginated listing, photos, search, recent,
trash, usage; folders, upload-url (+ multipart), {id}/complete,
{id}/download-url / preview-url / thumbnail-url; move / trash /
restore / delete-permanently; rename (PATCH /{id}); sharing, versions,
and activity.
Health — GET /health.
Deletion is soft:
DELETEmoves items to Trash; only/delete-permanently(and the scheduled purge) remove objects from storage.
Nimbus/
├── backend/ # FastAPI app — endpoints, services, repositories, storage, jobs
├── frontend/ # Next.js web app — landing, auth, dashboard
├── thumbnailer/ # S3-triggered thumbnail Lambda (Pillow)
├── infra/ # CloudFormation template
├── scripts/ # Lambda packaging, thumbnail backfill, Takeout migration
├── docs/ # architecture notes + screenshots
└── requirements.md # goals, cost model, and design rationale
Nimbus is open source and contributions are welcome. The best way to start is to read the codebase — it's organized by layer (see above) and every feature shipped as its own reviewed pull request, so the git history is a readable tour of how things were built.
- 🐛 Found a bug or have an idea? Open an issue.
- 🔧 Want to build something? Fork, branch, and open a PR — CI runs backend tests plus frontend type-checks and a production build on every PR.
- ⭐ Like the project? Star it on GitHub — it genuinely helps.
Licensed under the Apache License 2.0.
Made with ❤️ by Ankush Khakale · Your files, your cloud.