Voice & Asterisk / ARI / WebRTC Engineer

Location: Bengaluru (on-site / hybrid)
Experience: 3 to 7 years
Employment type: Full-time
Compensation: Competitive, benchmarked to experience.
About Us
Legodesk builds legal-powered credit and recovery infrastructure for India's banks, NBFCs, fintechs and ARCs. Our platform runs the full recovery lifecycle.
The Role
The Asterisk and its configuration, the SIP trunk and DID routing, the call-control layer that drives Asterisk through ARI, the browser softphone our agents work from, and the event pipeline our answer and outcome logic depends on.
What You will Own
The Asterisk platform
Deploy, configure, harden and operate Asterisk on AWS: chan_pjsip, dialplan, transports, codecs and module configuration
Patching, upgrades, monitoring and capacity planning. You decide when we upgrade and you carry out the upgrade
Design redundancy and failover for a production collections floor.
Secure the platform against toll fraud and SIP attack: firewalling, fail2ban or equivalent, credential hygiene, rate limits, and an SBC in front of the trunk if you judge it necessary
Know the concurrent-call ceiling of what you have built, and prove it with load testing rather than optimism
Call Control over ARI
Originate calls, bridge legs, manage conferences, start and stop recording, and hang up, all programmatically through the Asterisk REST Interface
Consume the ARI WebSocket event stream as the source of truth for call state: ringing, answered, hangup, failure and answering-machine detection
Make that stream reliable in production: reconnection, ordering, deduplication, missed-event reconciliation and state recovery after a drop. Our answer and outcome logic keys off these events
Keep call control behind a clean abstraction, so campaign, agent and disposition logic sits on a stable interface rather than being rewired every time the telephony layer changes
WebRTC agent softphone
Own the in-browser softphone for agents on Chrome: registration against chan_pjsip over WSS with DTLS-SRTP, per-agent endpoint credentials, and provisioning
Handle the hard parts: media permissions, device selection, NAT and ICE, reconnection on network flap, and call state sync between the softphone and the console
Diagnose and fix audio quality issues from the browser side: one-way audio, latency, jitter, packet loss and dropped legs
Own the agent experience end to end
SIP Trunk, DIDs and caller ID
Configure and own the SIP trunk on Asterisk: peering, registration or IP authentication, ports, codecs and inbound DID routing
Set outbound caller ID per call through ARI so our per-DID rotation for deliverability keeps working
Manage the DID inventory and CLI validity against what the carrier will accept
Own the carrier relationship on the technical side. You are the person who reads the SIP trace and tells the carrier what is actually wrong, with evidence
Recording and post-call pipeline
Control recording start and stop via ARI, and own where recordings are written, in what format, and how they reach our storage
Guarantee that every call has a retrievable recordings, campaign and agent, with no silent gaps
Keep the downstream transcription, AI summary and playback pipeline fed and healthy
Own storage lifecycle, retention and access control on recordings
What we are Looking for
Must Have
You have built and run Asterisk in production yourself: chan_pjsip, dialplan, trunk configuration, ChanSpy and ConfBridge, plus the patching, hardening and monitoring that come with owning the box
Hands-on production experience with ARI: originating and bridging calls programmatically and consuming the WebSocket event stream. Comparable FreeSWITCH ESL experience counts
Real WebRTC softphone work in the browser: SIP over WSS, DTLS-SRTP, and a library such as SIP.js or JsSIP wired into a live application
Working knowledge of SIP, RTP, codecs, NAT and ICE, and the ability to debug a call from trace to root cause
Experience configuring a carrier SIP trunk and securing a voice platform against toll fraud
Enough JavaScript to own the softphone in the browser, not just call an API from the backend
Experience with an outbound dialer or contact-centre platform at real volume.
Linux competence: systemd, Nginx, logs, performance tuning, and basic AWS (EC2, RDS, S3, VPC, security groups)
Ownership instinct. You are comfortable being accountable for a live production voice system
Strong Plus
You have built a browser-based agent calling experience on Asterisk that real agents used every day, not a proof of concept
Experience in collections, debt recovery, BPO or contact-centre tech; you know what a collections floor actually looks like
You have designed Asterisk high availability and failover for a live contact-centre floor, not just read about it
Redis, Celery or equivalent queueing
Speech tooling: ASR, TTS and diarization, especially for Indian languages
Familiarity with lending or BFSI compliance obligations
How to apply
Send your CV to ravi@legodesk.com
In a few lines, tell us about the dialing or voice system you have worked on: how many concurrent calls, what stack, what you personally built, and what broke.

Work arrangement
No