Senior AI Audio/Voice Deployment & MLOps Engineer

Team: Real-Time AI Voice Deepfake Detection Product Team

Location: Remote / Hybrid

Reports To: CTO

About the Role

Shama AI is seeking a Senior AI Audio/Voice Deployment & MLOps Engineer to take technical ownership of our production Voice AI platform. Our mission-critical product detects voice deepfakes in real-time across live telephony infrastructure, WebRTC streams, and on-premises appliances.

The role involves deploying and optimizing AI models and DSP algorithms for audio processing. It focuses on low-level programming for operating systems targeting deployment in local environments and resource-constrained devices while maintaining real-time performance. Working closely with the CTO, AI Research, and Product teams, you will architect and implement reliable AI deployment models for high-throughput systems while keeping low latency and memory constraints to analyze real customer conversations at scale.

Key Responsibilities

1. Model Optimization & MLOps Deployment

• Low-Latency & High-Throughput Inference: Convert, optimize, and deploy PyTorch models into high-throughput, low-latency production pipelines utilizing ONNX Runtime, CUDA, TensorRT, or NVIDIA Triton.
• 𝐇𝐚𝐫𝐝𝐰𝐚𝐫𝐞 𝐈𝐧𝐭𝐞𝐠𝐫𝐚𝐭𝐢𝐨𝐧: Deploy AI models on a variety of devices, from microcontrollers to gateways, and ensure they work efficiently with specific hardware accelerators (e.g., CPUs, DSPs, GPUs, NPUs).
• Cross-Architecture Compiling: Port high-level Python logic and signal processing algorithms into low-level, compiled C/C++ code optimized for both x86 and ARM architectures.
• Model Security & Packaging: Compile, containerize (Docker/Kubernetes), and securely encrypt models for cloud and on-premises appliance deployments.
• Infrastructure & MLOps: Design and maintain robust CI/CD pipelines, automated model evaluation workflows, and licensing systems.
• Observability: Monitor system-level performance metrics (CPU/GPU utilization, latency, memory usage) and manage auto-scaling, load balancing, and production incident debugging.

Required Qualifications

Technical Expertise

• 5+ Years of Industry Experience building, deploying, and operating production-grade Voice AI or real-time communication systems.
• Expert Programming Skills: Professional mastery of Python (including internals: CPython API, Cython, or PyBind11), alongside strong production experience in C/C++ and Golang.
• Deep Tech Stack Knowledge: Expert experience with asyncio, multithreading, Linux systems architecture, and distributed services.
• ML/Audio Deployment: Proven track record optimizing pipelines for live telephony or WebRTC streams using TensorRT, Triton, ONNX Runtime.
• Speech AI Models: 

Strong experience deploying speech-related AI models, including STT, TTS, ASR, and related technologies. 

Collaboration & Leadership

• Ownership Mindset: Ability to transition seamlessly from high-level architecture to hands-on implementation and rapid deployment.
• Cross-Functional Communication: Experience translating complex AI and infrastructure concepts into clear features for product teams and stakeholders.
•  𝐏𝐫𝐨𝐛𝐥𝐞𝐦-𝐬𝐨𝐥𝐯𝐢𝐧𝐠: The ability to approach problems from a user-centric perspective to create solutions that add real value. 
• 𝐓𝐞𝐚𝐦𝐰𝐨𝐫𝐤: The ability to work in a team environment.

Preferred Qualifications

• Audio DSP: Experience with digital audio signal processing, audio codecs, and transcoding platforms like FFmpeg.
• Hardware Familiarity: Basic understanding of electronics and hardware interfaces for local appliance integration.