Backend Engineer – Real-Time Video

Job description

About Stream Stream powers real-time Chat , Video , Activity Feeds , and AI Moderation for more than 1 billion end users across 4,000+ apps, from Strava and Bumble to eBay and Patreon. Our platform processes billions of API requests per month and supports applications with millions of concurrent users, while delivering highly reliable, low-latency services and a great developer experience. The role We're looking for a Staff Backend Engineer to join Stream on our mission of elevating the quality of apps for billions of users globally. You'll own the technical direction of Stream Video's media plane and the infrastructure it runs on. This is a full-time role, hybrid with 3 days a week in our Amsterdam office, or remote from anywhere in the EU. What you'll work on Stream Video runs on an SFU we built in Go on top of Pion, plus ingest and recording fleets, all on a global edge network we own across more than one cloud. Nobody at Stream has owned this media plane end to end before. Our central infrastructure team is tuned for request-based services, and video breaks in different ways: live sessions can't be dropped, bandwidth is the biggest cost, and placement is the main lever. That's why this is a Staff role. You'll set the direction for the video backend and its infrastructure with your own roadmap, reporting to our Director of Engineering with a direct line to our CTO. The Video SDK teams own the client side, and you'll work closely with them and with infrastructure. Part of the fleet is mid-move to Kubernetes. Job responsibilities: Architecture: write the architecture and roadmap for Stream Video's backend and media infrastructure, and decide what ships. Scale livestreaming: architect the SFU and media pipeline (WHIP, WHEP, RTMP, SRT, HLS, SVC), including the path from WebRTC to CDN-backed HLS for very large audiences. Own the fleet: reliability, autoscaling, placement, rollouts, observability and cost for long-lived sessions across clouds, with cost per gigabyte as a first-class metric. Gate production changes: define the validation that decides which changes, including AI-generated ones, reach live media, using behaviour tests on realistic networks, nightly telemetry scans and a public quality benchmark. Take the final escalation: handle the hardest quality, latency and production incidents, and work with enterprise customers on advanced integrations. Rai…