Senior SIP Engineer

About Telegent Want to use your mobile phone to set up call routing? Transfer a call to a colleague or an AI agent? Set up call filtering or a receptionist? All without an app, directly on the mobile network. ✅ Done. Now, want to do all of that without a SIM installed on a mobile device, or without a device at all? That's headless programmability: AI-native workflows on a real mobile number . Come help us build the mobile network for all agents, and the safe tech it can power .

Culture & pace We build fast, deliver fast, and ship fast, and what we ship is high-quality code. You'll move faster than you ever imagined and feel truly accomplished doing it. This may be some of the most meaningful work of your career: building technology that doesn't exist anywhere else. Expect to contribute to patents and verifiable work, and to learn more here than anywhere you've been.

Who you are A 200 OK is your love language, you've been ghosted by a missing ACK, and you know a 486 Busy Here isn't personal. One-way audio gives you NAT flashbacks, and you know it's always the SBC... until it's DNS 😆

Seriously , though, you:

• Have a deep understanding of SIP: registration and REGISTER/auth flows, NAT traversal, SDP/RTP , codecs, and SIP trunking. You'll get hands-on with SIP registration and WebRT C paths for integrations with UC and video platforms like Zoom, Microsoft Teams and Webex
• Know IMS core setup inside and out: P-CSCF, I-CSCF, S-CSCF, HSS, and how VoLTE and VoWiFi calls actually flow
• Have set up and configured SBCs: topology hiding, transcoding, security, and interop
• Know routing: call routing, number and L CR routing, failover and STIR/SHAKEN
• Have put AI voice agents on real calls: media streaming to AI, with latency kept lowenough that conversations feel natural
• Have mastery over agentic coding, including:

•

• Worktree management vs. multi-agent workflows, plus skills, hooks, and routines
• Flexibility across clients: Claude Code, Cursor , Codex, even custom harnesses like OpenClaw and Hermes
• Token optimization, plan mode, and structured approaches like SDD , TDD , SpecKit, and role-based development
• Knowing the difference between vibe coding and enterprise-grade agentic coding
• Geek out over paradigm-shifting approaches, like TypeSafe's Jev decision model routing an agent's steps, but don't let shiny object syndrome get in the way of shipping
• Know the end-to-end systems lifecycle and full-stack development, own your work end to end, and have deployed systems at scale with solid observability (when a call drops, you know exactly where and why)

You can't imagine coding by hand anymore. Your best friend might be an agent, and it probably speaks SIP . We don't judge. 🤖 You can't imagine a world without Wispr Flow , and you can't believe humans used to spend their time typing. Level: Senior

What you'll do

• Own call routing. Design and run how every call moves through our network, across our SBCs, media servers and carrier trunks, including keeping calls between our own numbers on our own network.
• Connect carriers and customers. Bring new SIP trunks and interconnects live in days, not weeks, and be the expert partners turn to when calls fail.
• Build programmable voice. Ship the features developers build on: call control APIs, webhooks, recording, IVR and real-time audio streaming to AI.
• Power our AI voice products. W ork with product on call screening, scam detection and voice agents that run on live calls.
• Retire legacy voice. Move traffic off older third-party voice platforms onto direct SIP trunks, with no downtime.
• Lead caller trust. Own STIR/SHAKEN, caller ID (CNAM) and recording consent, so our calls are signed, trusted and compliant.
• Make it reliable at scale. Redundant SBCs, routing by geography, capacity planning, monitoring and disaster recovery, as traffic grows many times over.
• Debug fast, then automate. Trace failed calls through SIP logs and packet captures, fix the cause, and turn every fix into a test or a runbook.
• Share what you know. Record decisions as ADRs, review code and make voice approachable for every engineer on the team.

The voice stack you will own:

Component & What it does

• SIP proxy / SBC - Call routing and SBCs at the edge, facing carriers and customers
• Media proxy - Media proxy that anchors call audio
• In-memory data store - Routing data looked up on every call
• Application server - Media and application server for call features: IVR, recording, bridging
• Programmable voice platform - Programmable voice layer behind our APIs and AI Voice apps
• Carrier SIP trucks - Direct links to our Tier-1 mobile partner and other carriers
• Speech and AI services - Speech-to-text, text-to-speech and AI agents on live calls
• Call monitoring and observability - Call tracing, call quality and alerting
• Cloud hosting and infrastructure as code - Multi-region hosting, defined as code

What we're looking for

• Deep SIP and RTP. You read SIP traces fluently and know SDP , codecs (G.711, G.729, AMR-WB, EVS), NAT traversal and media handling inside out.
• SIP servers and SBCs. Production experience with OpenSIPS or Kamailio, and FreeSWIT CH or Asterisk, at carrier scale.
• Carrier interconnects. SIP trunking with carriers, number routing and IP-based peering, plus the US rules that come with it: STIR/SHAKEN, CNAM and E911.
• IMS core. Hands-on P-CSCF, I-CSCF, S-CSCF and HSS setup, and how VoLTE andVoWiFi calls flow end to end.
• Troubleshooting. Wireshark, sngrep, SIPp, Homer or V oIPmonitor . You find the cause of a failed call, not just the symptom.
• Code and infrastructure. You write services and tooling in Python, Go or TypeScript, work with Redis and REST APIs, and run Linux, Terraform and cloud networking (Azure first).
• AI-native. You practice enterprise-grade agentic coding, not vibe coding, and you move easily across coding clients and harnesses. You arrive with your own setup of agents, skills and automations, and ideas on how to make ours better.
• Ownership and judgment. You take work from idea to production without hand-offs, raise problems early, and help keep the team pointed in the right direction.

Nice to have

• jambonz, or building CP aaS voice APIs (call control, webhooks, media streams)
• Running voice across several clouds and colocation sites
• Building to SOC 2 controls from the first line of code

How we work

• Learn, assess, benchmark, prepare solutions. Every responsibility runs this arc, done fast. Your first deliverable is your own assessment and 100-day plan.
• AI-first. Design and implementation are agent-driven across Claude Code, Cursor, Codex and multi-agent workflows, so one engineer moves at the pace of a team.
• You build it, you run it. No hand-offs to a separate operations team. You see the whole product end to end and ship it yourself .
• Small, reversible changes. Versioned, behind PRs with CI, with release tags and documentation that updates as the code does.
• Security from the start. New systems are built to SOC 2 controls, with architecture decisions recorded as ADRs.