Overview
Playfriends is a Web3 SocialFi platform where users join live voice and video lobbies hosted by creators, send virtual gifts, and book private sessions, all transacted in platform diamonds or on-chain tokens. It reached over 1M unique users with a 64% day-one retention rate.
My Role
Lead Backend & Cloud Engineer at MetaverseGo, owning the backend and cloud layer end to end.
Core API (NestJS)
Auth, payments, blockchain, user management.
Lobby Service (Node.js)
Room state, Agora RTC, gifting, bookings, real-time events.
AWS Infrastructure (CDK)
Multi-env IaC, ECS Fargate, networking, CI/CD.
Third-Party Integrations
Agora, Stripe, Coinbase, blockchain RPCs, Telegram, Discord.
Tech Stack
- Backend
- NestJSNode.jsTypeScript
- Cloud
- AWS CDKECS FargateLambdaCloudFrontAWS WAFSQS FIFORoute 53Amplify
- Real-time
- Agora RTCFirebase RTDB
- Web3
- Ethers.jsSolana Web3.jsTonApiWalletConnect
- Data
- MongoDB AtlasFirebase StorageS3
- Integrations
- StripeCoinbaseCoinMarketCapBrevoTelegrafDiscord.js
Platform Features
Live Voice & Video Lobbies
Host plus up to 8 co-hosts, audience presence, cloud recording.
Virtual Gifting
Real-time gifting with diamonds or on-chain tokens inside lobbies.
Private Bookings
Book co-hosts for 1:1 sessions, gaming, or companionship.
Real-Time Chat
In-lobby messaging with presence synced across connected clients.
Multi-Chain Wallets
Custodial wallets with EVM, Solana, TON, and Abstract L2 support.
$FRENS Native Token
Buyback-and-burn token with exclusive gifts and holder perks.
Fiat On-Ramps
Top-ups via Stripe and Coinbase with live rate conversion.
Cloud Recording
On-demand lobby recording piped to dedicated S3 storage.
Telegram & Discord Bots
Account linking, OTP verification, and activity notifications.
Impact
At 1M+ users with live payments, we cleared attacks, payment abuse, and inefficient data paths—then drove the cost work below.
DDoS and edge defense
Traffic spikes toward the app. Edge filtering + auto-scaling kept us online while rules tightened.
Abuse on payments and top-ups
Probes on money paths. Stricter limits on those routes; one credit per on-chain transaction—no double-credits.
Database slowdowns at scale
Slow queries under load. Short-term scale-up, then indexes and query fixes so we could run smaller again.
Realtime data doing too much work
Lobby sync moved excess data. Narrower writes and subscriptions; less waste, same UX.
Outcomes
Led a cost optimization effort across AWS, MongoDB Atlas, Agora, and Firebase that reduced total monthly infrastructure and third-party spend by roughly 65%, with no regressions to the user experience.