Ryan S.

I help contact centers and phone-heavy businesses design, deploy, and harden production voice AI systems that work in the real world. The voice agents I've built have handled millions of live calls. The scoring and QA systems I've built have evaluated millions more. I've spent 20 years building phone systems and contact center infrastructure. Twenty years ago, that meant on-prem PBX integrations. Ten years ago, it meant cloud contact centers on Twilio. Today, it means production AI voice agents, call scoring pipelines, and the backend systems that make them reliable at scale. I'm the founder and CTO of Automatdo, a voice AI platform for contact centers, and I still write production code every day. Before that, I was CTO and partner at HelloGym, where I built the SIP telephony network and contact center infrastructure that processed millions of calls. In 2020, I built the platform that a global pharmaceutical manufacturer used to stand up their COVID-19 response contact center in six weeks on Twilio and serverless infrastructure. The same stack was later deployed for other Fortune 500 brands. I spent 18 months as the primary builder on the comms platform at Veho (now $1B+). That combination matters: I understand voice AI from the LLM layer down through the SIP stack, not just the parts that are trendy this year. I work with teams that have real phone volume, real operational constraints, and no interest in learning painful lessons in production. That includes contact centers, service businesses, and companies evaluating how voice AI should fit into customer support, qualification, scheduling, verification, and follow-up workflows. Typical engagements fall into four buckets: - Building production voice AI agents end to end, for inbound qualification, outbound follow-up, scheduling, TPV verification, and full customer service flows. - Auditing architecture, vendors, and call flows before teams commit budget to the wrong stack. - Improving contact center operations with AI scoring, QA, and evaluation systems. - Solving Twilio and CPaaS problems across Voice, Flex, TaskRouter, Studio, SIP, number operations, branded calling, and compliance. The stack varies by use case, but the hard parts stay the same: latency, interruption handling, handoff logic, bad transcripts, edge cases, and making sure the system still works when the call does not go to plan. My core stack is Python, FastAPI, Django, Twilio, PostgreSQL, and AWS, combined with modern LLM, TTS, and STT APIs. I'm comfortable across Twilio Voice, Vonage, ElevenLabs, Deepgram, Whisper, and self-hosted components when the tradeoffs call for them. I'm direct and implementation-minded. I'll tell you when an architecture is overcomplicated, when a vendor choice is wrong, or when voice AI is the wrong solution for a workflow. I ship working systems, not slideware, and I write code your team can own after the engagement ends.