AI Voice Agent Developer Needed — Twilio/Vapi/Retell, LLM & Calendar Integration

AI Voice Agent / Full-Stack AI Developer — Existing Product Enhancement We are looking for an experienced AI Voice Agent / Full-Stack AI Developer to help improve, optimize, and expand an existing AI-powered voice call agent already under development. This is an existing product owned entirely by us. We already have a software engineer who will remain the Lead Architect / Technical Lead. The selected developer will work alongside our engineer to review the existing codebase, resolve technical issues, optimize performance, and add remaining functionality. We are not looking to rebuild the product from scratch. Major architectural changes or rewrites must be discussed with and approved by our Lead Engineer. Main Priority: Voice Speed & Natural Conversation One of our biggest current challenges is voice-agent response speed and conversational flow. We need someone with hands-on experience diagnosing and improving end-to-end conversational latency, including: * Speech-to-Text latency * LLM time-to-first-token * Text-to-Speech latency * Streaming STT → LLM → TTS * Voice Activity Detection / turn detection * Barge-in and interruption handling * Natural conversation pacing * API/tool-call and calendar lookup latency * Prompt/context and backend optimization For simple conversational responses, our goal is for the agent to begin responding as quickly as technically practical, ideally around 1 second or less where achievable. We specifically want someone who can measure where latency is occurring in the existing system rather than simply replacing models or providers without diagnosing the underlying issue. Core Functionality The voice agent should support: * Incoming customer calls * Natural conversational speech * Maintaining context * Appointment booking * Appointment cancellation * Appointment rescheduling * Real-time availability checking * Service/pricing information * Customer information collection * Human escalation/transfer when appropriate * Call logging and summaries The system may integrate with scheduling platforms such as Google Calendar, Microsoft Outlook, Calendly, Square, Acuity, or similar systems. Zero-Hallucination Requirement The AI must never invent: * Services * Prices * Appointment availability * Business hours * Policies * Service details * Booking/cancellation/rescheduling confirmations Service information must come from configured business data. Appointment availability must come from the actual scheduling system, and the AI must only confirm an action after the backend/API confirms that it succeeded. Reliability & Double-Booking Prevention The system should safely handle API failures, timeouts, telephony failures, LLM failures, unavailable appointments, concurrent calls, and failed booking attempts. The booking implementation must account for race conditions and double bookings, including verifying availability as close as possible to the final booking operation. Preferred Experience Relevant experience may include: Voice / Telephony: Twilio, Telnyx, Vapi, Retell AI, SIP/VoIP, WebSockets, streaming audio, STT/TTS, VAD, barge-in. AI / LLM: OpenAI APIs, Anthropic Claude, function/tool calling, structured outputs, agent workflows, prompt engineering, context management, guardrails. Backend: Node.js/TypeScript, Python/FastAPI, REST APIs, webhooks, PostgreSQL, Redis, authentication, logging, async/real-time systems. The developer must be comfortable working within our existing technology stack and under our existing Lead Architect. Ownership — Critical Requirement We will retain 100% ownership of the existing product and all custom project-specific work created for this project. The developer will not receive ownership, equity, licensing rights, or intellectual-property rights in the product. Upon payment for applicable work, all custom project-specific source code, prompts, workflows, integrations, business logic, database structures, dashboard/frontend code, documentation, tests, and deployment/configuration files will belong exclusively to us. All project code must be committed to repositories controlled by us. Production accounts and infrastructure should also be under accounts controlled by us whenever practical. Any pre-existing proprietary libraries, frameworks, code, or components must be disclosed before implementation. Working With Our Engineer Our existing engineer will remain the Lead Architect / Technical Lead. You must be comfortable: * Joining an existing codebase * Following an established architecture * Collaborating through Git * Creating clean pull requests * Receiving code review * Explaining technical decisions * Improving existing components instead of unnecessarily replacing them Proposal Requirements Please include: 1. Examples of AI voice agents you have built and your specific role. 2. Voice/telephony platforms you have worked with. 3. Your experience optimizing voice-agent response time. 4. Approximate end-to-end latency you have achieved previously. 5. How you would diagnose latency in an existing system. 6. How you handle barge-in/interruption. 7. Experience with calendar/appointment APIs. 8. How you prevent double bookings/race conditions. 9. Experience working inside another engineer’s existing codebase. 10. Estimated timeline after reviewing the current system. 11. Preferred pricing structure: hourly or milestones. 12. Confirmation that we will own 100% of custom project-specific work. 13. Confirmation that all project code will be committed to our repositories. 14. Disclosure of any proprietary/pre-existing components you intend to use. Please begin your proposal with “VOICE AGENT” so we know you reviewed the requirements. A detailed technical scope document is attached and should be reviewed before submitting a proposal. I also created the full technical-scope Word attachment from your engineer’s updated version, including the stronger ownership language: