Senior Software Engineer, Real-Time Communications
- Market p25–p75
- $167k–$213k
- Tin Can p25–p75
- $149k–$166kn=4
- Opened · 30d
- +0
- Closed · 30d
- −3
- Opened · 7d
- +1,3388% of open
- All software engineering
- 75,094open
About Tin Can
Tin Can is building a safer, simpler way for kids to connect — without smartphones. We’re creating screen-free, delightful devices and services that let families call the people who matter most, free from the noise of today’s digital world.
We’re building a bold, authentic, nostalgic, and kinda quirky brand that resonates with folks who want something simpler & better for their kids than the tech-infused lives we’re currently living (and who have a sense of humor about it). As we gear up to scale to thousands of families, we’re ready to bring on an engineer to help make it happen.
The Role
A call either connects or it doesn’t — there’s no partial credit when a kid picks up a Tin Can to reach Grandma, or to dial 911. We’re hiring a Senior Software Engineer, Real-Time Communications to own the path a kid’s voice takes: how a call routes, how it sounds, and how we know it worked. That path spans two generations — the SIP plane carrying every call today, and the WebRTC handset leg we’re building next.
Today that’s Kamailio and FreeSWITCH on AWS, rtpengine at the edge, carrier trunks, and a TypeScript control plane that gates every call. Next is WebRTC on the handset, bridged into that SIP core, because 911 and the phone network will always speak SIP. It’s a deeply hands-on role: routing logic, media paths, and packet captures every day, not a management chain. You’ll report to the Head of Engineering and own the call itself. When a registration flaps at 3am, you go from trace to root cause to fix.
What You’ll Do
Own the production voice path end to end — SIP signaling, media, and the carrier trunks behind every outside call
Design how WebRTC handsets join that path: signaling, TURN, and the WebRTC-to-SIP bridge that keeps 911 and the phone network reachable
Write the routing logic and dialplans behind every call — Kamailio, FreeSWITCH, and our TypeScript control plane — so adding a market is configuration, not code
Be the reviewer of record for call-path changes, and author the Sipfront and SIPp scenarios that gate them before they ship
Produce the packet-level truth about call quality — MOS, jitter, loss, setup time — and make it trustworthy enough to alarm on
Model and load-test call capacity — concurrent calls, calls per second, trunk headroom — against our holiday peak
Extend calling into new countries, where codecs, carriers, and emergency-calling rules change market to market
Harden the path: SIP over TLS, SRTP, rate limiting, toll-fraud prevention
Make this stack legible to every engineer here: the documentation, runbooks, and reviews
What We’re Looking For
5+ years building and operating real-time voice or video systems in production, including a stretch where the call path was yours: it broke, you fixed it, and you changed what you measured
Deep in SIP and RTP on a production stack — Kamailio, OpenSIPS, FreeSWITCH, Asterisk, Janus, mediasoup, LiveKit, or a CPaaS platform
You can hold your own on registration, dialog state, NAT traversal, codecs and jitter buffers, and you’ve taken a live production call from a packet capture to root cause
At home in production code as well as config; both reviewed, tested, reproducible
Fluency using AI as a multiplier across diagnosis, analysis, and testing; this is how we work, not a nice-to-have
A clear communicator who writes things down, and a collaborative, low-ego approach — everyone here sweeps the floor
Based in Seattle and ready to work from our office Monday through Thursday; this isn’t a remote role
Bonus: Kamailio, FreeSWITCH, or rtpengine; WebRTC internals and bridging WebRTC to SIP; Homer/HEP; E911 or carrier interconnect; STIR/SHAKEN
Why Join Tin Can
We’re building tech that protects childhood: We’re on a quest to give kids a more analog childhood — one with real conversations, real connection, and way less screen time. No doomscrolling, data mining, or dopamine traps — just a simpler, better way for kids to stay in touch. This is a rare opportunity to build something that truly matters.