Website:
seeclearstudio.com
Job details:
Location: Remote
Position type: Contract
Experience: 4+ years
About the role
A contract engineer who independently builds and runs our configurable Voice SDR module: a context-aware voice agent that receives an Engagement Assignment, understands the business and prospect context, conducts adaptive conversations, and returns structured outcomes and intelligence.
Responsibilities
• Build the Voice SDR module end to end, including its four internal engines: Context & Knowledge, Conversation, Decision & Action, and Memory & Intelligence.
• Design the agent as a configurable, company - agnostic capability; use the initial IT - services sales scenario as a demonstration tenant, not as an architectural limitation.
• Implement Engagement Assignment ingestion and a Context Assembler that uses company profile, prospect intelligence, prior emails/social interactions, call purpose, commitments, permissions and engagement history.
• Build dynamic conversation-state management and decision-making so the SDR interprets meaning, adapts to unexpected responses, qualifies appropriately and selects the best outcome - not merely executes scripts or forces meetings.
• Integrate a real-time Voice Runtime and telephony provider (Vapi, Retell, LiveKit, Pipecat or equivalent), including speech-to-text, text-to-speech, custom tenant voice, turn-taking, interruptions and call lifecycle handling.
• Implement controlled knowledge retrieval, Tool/Action Gateway, calendar booking, follow-up actions and Human Escalation; never invent facts, pricing, capabilities or commitments.
• Implement session memory and persistent conversation intelligence: distinguish explicit facts from inferences, capture requirements, objections, referrals and commitments, and generate structured call outcomes and next-action recommendations.
• Define and document versioned inbound/outbound contracts, APIs, events and webhooks for a separately developed integration adapter; support stable IDs, idempotency, retries, tenant isolation and auditability.
• Keep a central system of record. Do not rebuild Prospect Intelligence, canonical prospect records, channel selection or platform-wide workflow orchestration inside the Voice SDR.
• Test and guardrail the system through simulated and authorised real-call scenarios, including latency, interruptions, voicemail, silence, stale contacts, gatekeepers, unexpected responses, opt-outs and future-call continuity.
• Deliver architecture, working implementation, automated behavioural tests, deployment/configuration guidance, observability and an end-to-end demonstration.
Mandatory skills
• At least one real-time voice agent shipped to real users (Vapi, Retell, LiveKit, Pipecat or similar), which they can demonstrate and explain technically.
• 4+ years of backend engineering in Python or Node.js: APIs, webhooks, queues, PostgreSQL or similar.
• Hands-on LLM engineering: orchestration, prompt design, tool calling, structured outputs and reliable state management.
• Practical knowledge of speech-to-text, text-to-speech, turn-taking, interruption handling and real-time latency.
• Experience building context-aware conversational systems, retrieval/knowledge workflows, memory and automated AI behavioural testing.
• Systematic engineering practices, security and multi-tenant data handling; daily use of AI coding tools.
• Ability to own the build independently, work directly with the team and deliver a production-oriented module against agreed interfaces.
Preferred skills
• Sales/SDR, CRM or engagement automation experience.
• Telephony experience (Twilio, Plivo or Exotel) and calendar integrations (Google Calendar, Calendly or Cal.com).
• Experience with real-time distributed systems, RAG/knowledge bases, event-driven architecture and production observability.
Click on Apply to know more.