Senior AI Infrastructure Engineer - Voice AI

Job Overview

• We are looking for a Senior Voice Infrastructure Engineer who can own and scale the infrastructure behind real-time Voice AI systems. This role is focused on serving speech-to-text, language models, and text-to-speech models under strict latency, reliability, and cost requirements.
• The ideal candidate should have strong production infrastructure experience, hands-on exposure to GPU inference serving, Kubernetes, cloud infrastructure, and infrastructure-as-code. This is a high-ownership engineering role where the person will be expected to make architectural decisions, optimize GPU usage, improve reliability, and reduce cost-per-minute at scale.
• The role is best suited for someone who has worked on low-latency systems, AI model serving, real-time platforms, voice/audio infrastructure, or GPU-heavy production environments.

Key Responsibilities

• Own infrastructure for serving STT, LLM, and TTS models in production.
• Manage GPU inference workloads under strict latency and reliability requirements.
• Design and operate scalable GPU serving infrastructure for real-time voice conversations.
• Optimize cost-per-minute by improving GPU utilization, scaling strategy, and model serving efficiency.
• Work on Kubernetes-based infrastructure, cloud deployments, and infrastructure-as-code.
• Build and maintain observability across latency, throughput, GPU usage, uptime, errors, and cost.
• Implement warm pools, autoscaling, scale-to-zero, and capacity planning strategies where required.
• Improve reliability, incident response, and on-call readiness for production systems.
• Ensure security and compliance for a multi-tenant platform handling voice data and PII.
• Work closely with ML, product, and engineering teams to support fast and reliable model deployment.
• Make major architectural decisions independently and execute them quickly.

Ideal Candidate Profile

• 5+ years of experience running production infrastructure.
• Strong hands-on experience with Kubernetes, cloud infrastructure, and infrastructure-as-code.
• Experience with AWS is preferred.
• Hands-on experience with GPU inference serving and managing GPU workloads in production.
• Experience with inference serving tools such as vLLM, Triton, TensorRT-LLM, or similar platforms.
• Strong understanding of low-latency system design, autoscaling, observability, and reliability engineering.
• Experience optimizing infrastructure cost, GPU utilization, and production performance.
• Ability to own complex systems end to end with minimal supervision.
• Strong debugging and incident-management skills.
• Comfortable working in a fast-moving startup environment with high ownership.

Good to Have

• Experience with real-time voice, telephony, streaming, audio, or communication infrastructure.
• Experience supporting STT, TTS, LLM, or Voice AI model serving.
• Experience as an early infrastructure hire or founding-era engineer.
• Strong FinOps mindset and experience reducing cloud or GPU costs at scale.
• Experience with multi-region infrastructure and distributed systems.
• Interest in voice AI, speech technology, and emerging markets.

You Are a Good Fit If

• You have managed production infrastructure, not just built proof-of-concepts.
• You have worked with GPU serving or AI inference workloads.
• You understand latency, reliability, and cost as core infrastructure metrics.
• You can design scalable systems and operate them under real production pressure.
• You are comfortable owning infrastructure decisions independently.
• You can work closely with ML teams to serve models reliably and efficiently.

You May Not Be a Fit If

• You have only worked on traditional web infrastructure without AI or GPU workloads.
• You have not managed production systems at scale.
• You are not comfortable with on-call ownership or production incidents.
• You have limited experience with Kubernetes or cloud infrastructure.
• You are looking for a pure DevOps support role without architectural ownership.