AI Engineer

• *AI Engineer — Voice \& Language**

4\+ Years · Full\-Time· Jaipur **About the Role**

We're hiring a senior AI Engineer to design, build, and ship production AI systems — with strong emphasis on Voice AI. You'll own the full lifecycle: architecture, training, deployment, and monitoring across language and voice modalities. **What You'll Do*** **LLM \& GenAI:** Fine\-tune and deploy LLMs; build RAG pipelines and agentic workflows (LangChain, LlamaIndex).

• **Voice Pipelines:** Architect real\-time ASR LLM TTS pipelines with \<300 ms latency
• **Voice Agents:** Build production voice agents with turn\-taking, barge\-in handling, and emotion\-aware dialogue.
• **Speech Fine\-Tuning:** Adapt ASR/TTS models for domain\-specific accents, terminology, and speaking styles.
• **MLOps:** Build reproducible ML pipelines (Kubeflow / MLflow); maintain CI/CD, monitoring, and model versioning.
• **Inference Optimization:** Apply quantization (GGUF, GPTQ), distillation, and hardware\-aware inference (TensorRT, vLLM) to cut cost and latency.
• **APIs \& Services:** Ship high\-performance inference APIs in Python (FastAPI) or Go on Kubernetes.
• **Data \& Evaluation:** Curate text \+ speech corpora; define eval harnesses covering WER, MOS, latency P95, and safety.

• *Requirements** **Must\-Have**

• 4\+ yrs ML/software engineering; 2\+ yrs on production AI systems
• Strong Python; PyTorch or TensorFlow
• LLM fine\-tuning: LoRA / QLoRA / PEFT

End\-to\-end ML pipeline experience (train* serve)

• Cloud (AWS / GCP / Azure) \+ Docker / Kubernetes
• ASR \& TTS integration in real\-time streaming systems
• VAD, noise suppression, and barge\-in handling
• Telephony APIs (Twilio, Vonage) or WebRTC experience

• *Nice\-to\-Have**

• Whisper / wav2vec fine\-tuning for domain adaptation
• Audio\-language models (AudioPaLM, Qwen\-Audio, Gemini Audio)
• Speaker diarization (pyannote.audio) or voice biometrics
• Prosody control, SSML, expressive TTS synthesis
• Multilingual ASR/TTS and code\-switching pipelines
• RLHF / Constitutional AI alignment
• Vector DBs (Pinecone, Weaviate, pgvector)
• Open\-source contributions or published research

• *Tech Stack** **Core****Python** **PyTorch** **FastAPI / Go** **Kubernetes** **MLflow** **LLM \& GenAI****OpenAI / HuggingFace** **LangChain** **LlamaIndex** **vLLM** **RAG / Agents** **️ Voice AI****STT** **TTS** **WebRTC / WebSockets** **pyannote.audio** **Twilio / Vonage** **️ Audio Processing****librosa / FFmpeg** **Silero VAD** **openWakeWord** **SSML / Prosody** **AEC / Noise Suppression**