Agentic Engineer - Voice AI
At a glance
Mid-level AI Engineering role at WATI.io. Shenzhen.
Pay not stated AI Engineering salary
aimlroles summary, based on the employer's posting.
What you'll do
- Build the real-time voice path from call audio transport through model processing to spoken responses
- Connect live AI models and speech-recognition, language-model and speech-generation stages into voice conversations
- Maintain audio routing, codec handling, transport protocols and voice-activity detection
- Add agent functions for tool use, context handling and conversations that span multiple turns
- Reduce response delays and strengthen the reliability and scale of WhatsApp voice infrastructure
What you bring
- At least three years in software engineering, with solid backend development in Go or Python preferred
- Hands-on familiarity with real-time communications, such as WebRTC, RTP/RTCP, audio codecs or media servers
- Working knowledge of AI model connections, including APIs, tool use, prompt design or agent coordination
- Experience or strong interest in speech recognition, speech synthesis, voice detection or audio processing
- Understanding of distributed services and cloud-native systems, plus PostgreSQL, Redis and pub/sub messaging
Who this fits
This mid-level, onsite role in Shenzhen suits a backend engineer interested in real-time communications and voice AI. You'll work across product and platform teams and debug systems where response time matters.
From the employer
About Wati
Started as a WhatsApp team inbox in 2020, Wati has evolved into an AI-powered customer engagement platform that goes beyond a single channel. Designed for businesses that sell, support, and grow through conversations, Wati observes customer intent in real time, decides the next best revenue action, and executes it across marketing, sales, and support — on WhatsApp, Instagram, Facebook, TikTok, SMS, and more.
Trusted by over 16,000 customers across 190+ countries, Wati simplifies complex operations and business conversations with a unified inbox, no-code automation, and our intelligent AI layer, Astra.
Proudly backed by Tiger Global, Sequoia Capital, DST Global, and Shopify, and recognised as a Premium Partner of Meta and Google.
About the Role
We are looking for an Agentic Engineer – Voice AI to build and scale Wati's real-time voice AI capabilities on WhatsApp.
In this role, you will develop the systems that allow AI agents to listen, think, and speak in real time over WhatsApp voice calls. This includes building and optimizing the real-time media pipeline (WebRTC, LiveKit), integrating frontier AI models (OpenAI Realtime API, Google Gemini Live), and engineering the cascade architecture that connects speech recognition, language models, and speech synthesis into a seamless conversational experience.
You will work on latency-critical infrastructure where every millisecond matters — from audio transport and voice activity detection to model inference and text-to-speech delivery. You will also contribute to the broader AI agent stack, including tool calling, context management, and multi-turn conversation orchestration.
This role sits at the intersection of real-time communication systems, AI model integration, and conversational voice experiences.
What You Will Do
• Design, build, and optimize real-time voice AI pipelines — from WebRTC media transport to LLM inference and speech synthesis
• Integrate and orchestrate frontier AI models including OpenAI Realtime API, Google Gemini multimodal live, and cascade architectures (ASR → LLM → TTS)
• Build and maintain the media infrastructure: LiveKit-based audio routing, Opus codec handling, RTP/RTCP transport, and voice activity detection
• Develop agent capabilities for voice interactions — tool calling, function execution, context engineering, and multi-turn conversation management
• Optimize end-to-end latency across the voice pipeline, from audio capture to AI response playback
• Collaborate with product and platform teams to deliver production-grade voice AI experiences on WhatsApp
• Ensure reliability, performance, and scalability of voice AI infrastructure serving customers across 190+ countries
Requirements
• 3+ years of software engineering experience, with strong backend development skills (Go or Python preferred)
• Experience with real-time communication technologies: WebRTC, RTP/RTCP, audio codecs, or media server infrastructure
• Familiarity with AI/LLM integration — model APIs, tool calling, prompt engineering, or agent orchestration
• Experience with or strong interest in speech technologies: ASR, TTS, voice activity detection, or audio processing pipelines
• Understanding of distributed systems, microservices, and cloud-native architectures (GCP preferred)
• Comfortable working with PostgreSQL, Redis, and pub/sub messaging systems
• Strong problem-solving ability and ability to work in fast-paced, ambiguity-rich environments
• Ability to debug complex, latency-sensitive systems by reading code, traces, and real-time metrics
Nice to Have
• Hands-on experience with OpenAI Realtime API, Gemini multimodal live, or similar real-time AI model APIs
• Experience with AI agent frameworks (Dify, LangChain, CrewAI, etc.)
• Familiarity with MCP (Model Context Protocol) or other agent integration standards
• Contributions to open-source projects in the AI or real-time communication space
Behavioural Expectations
• Strong ownership and bias for action in a fast-moving environment
• Proactive, self-driven, and comfortable working across teams to drive outcomes
• AI-native mindset, actively using AI tools in daily engineering workflow and experimenting with agent frameworks and emerging technologies
• Curiosity-driven — eager to push the boundaries of what voice AI can do in real business contexts