Reconnect in 6 seconds instead of 30
- Problem.
- After a network drop, it took a participant about 30 seconds to get back into the call.
- What I did.
- Reworked connection recovery in the client WebRTC library.
6 s instead of 30 — five times faster
Frontend Engineer · 6 years · 4 years of WebRTC
TypeScript and React. For the last four years I’ve worked on video conferencing: WebRTC, SFU and MCU, HLS, telemetry. Today I build the client WebRTC library at MTS Link.
Reconnect 30 → 6 s · MTS Link
model · after a network drop
Media Core team, the client WebRTC library, since July 2024.
No release
SFU + MCU
Plus the engineering groundwork: ~110 tests, linting and git hooks, unified error handling, integration docs.
6 s instead of 30 — five times faster
A top recurring Sentry error — gone
Built from scratch and shipped to production
Six years of commercial development, four of them in video conferencing.
Jul 2024 – present
Client WebRTC library. One of its two core developers.
Mar 2022 – Jul 2024
VKS.FSO — a video conferencing web client on top of the Vinteo MCU, built from scratch and shipped to production.
Nov 2020 – Feb 2022
The web interface for AdminBox, a hardware vulnerability scanner, built from scratch.
the platform · since 3 September 2026
My own development platform. Products on it are built by Claude Code agents from specs: WHEN/THEN requirements, agent roles, automated checks, batched releases.
Five products for clients. The demos sit behind a login — message me and I’ll send you login details.
A role, a product or a question about the platform — reach out however suits you.
Telegram is the fastest way to reach me.