Arpit Mishra
Fullstack AI Developer
Socials
Low-latency media, WebRTC and live streaming architecture
Every streaming architecture is really a trade-off between latency and audience scale — here's how HLS, DASH, and WebRTC sit on that spectrum, and how to pick between them.
How a stream gets from a streamer's webcam to a million viewers' screens — capture, ingest, transcode, and adaptive-bitrate delivery.
Almost nobody has a routable public IP anymore — here's how two peers behind separate NATs still manage to find each other and exchange media directly.
A look under the hood of a Selective Forwarding Unit — producers, consumers, simulcast, and how Mediasoup-style SFUs scale past a single machine.
Why every group video product eventually has to choose between mesh, MCU, and SFU topologies — and why almost all of them land on SFU.
Sockets always relay through a server; WebRTC tries to skip the server entirely. Here's what that trade actually costs you.
TCP guarantees everything arrives, in order, at the cost of waiting for it. Real-time media can't afford to wait — here's why WebRTC bets on UDP instead.
