Ali Abbas Solangi

Why

• I wanted a vault I could host myself and use with people I trust, instead of dropping files into someone else’s cloud.
• I wanted conversion inside the product, so I could change a file’s format without leaving the vault, and the option to compress images on upload.
• I wanted shared folders so friends could work in the same place, and timed links so a file could go out without creating another account.

What I did

• I built SafeKeeping and shipped it at safekeeping.lol: a Go API and a single-page vault on Railway.
• Personal files are encrypted in the browser before they upload. Shared folders, 24-hour links, in-vault conversion, and optional image compress all live in the same product.
• New accounts wait for my approval before they can upload. I set up the domain, DNS, and production email through Resend.

What I learned

• Go — a full API, sessions, and file handling in one process, not a hello-world service.
• Browser encryption with AES-GCM, and why shared folders cannot use the same model if more than one person needs to open the files.
• Production email when SMTP is blocked: Resend over HTTPS, plus the DNS that makes those messages actually send.

Why

Me and my friends were always playing social games. We hated it when a game made everyone connect on different phones for something that did not need that. And every game we actually played sat in its own app, even though they were all party games.

What I did

I took it into my own hands and built the party-game web app I wanted. Spyfall, Mafia, Impostor, and more live in one place. I designed the brand in Photoshop and shipped the whole thing myself — Cursor, GitHub, Vercel, Supabase. We use it day to day.

What I learned

• Owning a product from logo to a URL people open taught me more than any tutorial — especially the bugs only real players find
• How to grow something with no budget: friends, societies, and community chats, not ads
• Shipping full-stack on Vercel and Supabase, moving faster in Cursor, and making a game simple enough that a group will actually start it