Description
Summary • Design and build a custom voice agent pipeline (STT → dialogue engine/LLM → TTS → telephony) optimized for low latency and natural conversation. • Set up and harden a self-hosted server environment where all call recordings, transcripts, and patient interaction data are stored securely — not routed through or retained by third-party SaaS platforms. • Implement encryption in transit and at rest, role-based access control, audit logging, and secure API/credential management across the stack. • Integrate telephony (e.g., Exotel, Plivo, or equivalent) for inbound/outbound calling, and manage the SIP/voice infrastructure. • Architect the platform to be multi-tenant — our clinics will each have isolated data with logically separated access, since this will grow into a subscription (SaaS) product serving many clinics, not a single deployment. • Build for India’s DPDP Act requirements: data residency (India-based hosting), consent logging, retention/deletion policies, and breach-readiness. • Continuously optimize for cost and latency — we are targeting an all-in cost per minute that keeps the product commercially viable at clinic-level pricing. • Proactively suggest architecture, tooling, or product improvements. We want a technical partner who brings ideas, not just someone who executes tickets. • Document the system clearly enough that it can be maintained, audited, or handed off if needed — full IP and code ownership remains us at all times.