Voice Bot SaaS Platform for Hebrew Conversational AI

Hi I am currently moving forward with a new project and am looking for a strong Full-Stack Developer to lead our MVP development. Below is a summary of what we are building so you can see if it aligns with your tech stack and experience. 1. What is the Product? A multi-tenant SaaS platform that allows small and medium-sized businesses to deploy an AI-driven, automated voice agent (phone bot). The bot handles customer service, manages orders, schedules appointments, and answers FAQs—operating 24/7 in fluent Hebrew. The platform consists of two primary interfaces: Admin Panel: For global client management, phone number provisioning, and bot creation using built-in templates (e.g., pizzeria, dentist, or fully customized bots). Client Portal: A dedicated, isolated dashboard for each business to view call logs, read transcripts, and manage orders or appointments generated during the calls. 2. Technical Architecture (The Stack) The core of the system is built on real-time communication and complex voice-to-AI integrations: Backend: Node.js + TypeScript (utilizing a persistent server like Fastify/Express rather than Serverless due to active WebSocket state management). Frontend: React + TypeScript + Tailwind CSS (with shadcn/ui). Database & Cache: PostgreSQL (via Prisma ORM) + Redis to manage the real-time state of active phone sessions. Telephony & Telecommunication: Twilio Voice leveraging the ConversationRelay component for end-to-end media streaming over WebSockets. AI & Voice Engine: ElevenLabs Flash v2.5 for low-latency Hebrew Text-to-Speech (TTS). State-of-the-art LLMs (Claude / GPT-4) running in a hybrid configuration: a rigid Q&A tree paired with dynamic AI fallback through Tool/Function Calling. 3. Key Development Phases (MVP Roadmap) Phase 0: Infrastructure setup, database schema design (ensuring strict multi-tenant isolation), and JWT-based authentication. Phase 1 (Core Engine): The conversation pipeline — establishing the Twilio WebSocket connection, integrating ElevenLabs, and building the hybrid conversation manager. The primary success metric here is achieving a strict response latency target of under 2 seconds. Phase 2: Admin Panel & Bot Builder (developing a graphical wizard for configuring workflows and customizing system prompts). Phase 3: Client Portal (delivering the business-facing dashboard, call history logs, transcriptions, and order status management). (Future Phases: Native integrations with Google Calendar and external CRMs via LLM Tool Calling). 4. Who Are We Looking For? We are seeking an autonomous developer capable of spinning up architecture from scratch, with a particular focus on: Proven, hands-on experience with WebSockets and real-time backend systems. Deep familiarity with AI APIs (LLMs, Prompt Engineering, and Tool/Function Calling). Strong Advantage: Prior experience with Twilio Voice (or similar VoIP frameworks) and STT/TTS integrations. Performance-driven mindset: Latency optimization is our absolute priority. Estimated MVP Timeline: Approximately 10–12 weeks of development. If this aligns perfectly with your experience and you have immediate or near-term availability, I would love to schedule a brief call to dive deeper into the details. Best regards, Harel s