Production ML/AI Engineer for ASR & LLM Solutions

Verbit, Inc.; United States; Remote; USD 140,000 - 190,000; Full time

Job summary

Verbit, Inc. is seeking a skilled ML/AI Engineer to build, deploy, and serve ML models in production, focusing on ASR-related tasks and scalable speech solutions. The role emphasizes learning in the speech domain and improving inference performance.

The ideal candidate has 4+ years in ML, strong Python, and experience with PyTorch/TensorFlow, Docker, and cloud-based MLOps. You will work to optimize latency and integrate advanced AI features across Verbit’s platform.

Qualifications

• 4+ years of experience in machine learning/deep learning.
• Strong Python and Linux skills.
• Experience with PyTorch or TensorFlow and training workflows.
• Experience with MLOps pipelines and production model serving.
• Experience with Docker, gRPC, and cloud-based serverless/micro-services.
• Experience with real-time/low-latency streaming systems.

Responsibilities

• Build, deploy, and serve ML models in production, optimizing latency and GPU usage.
• Extend speech solutions with 3rd-party ASR integrations and self-hosted models.
• Evaluate ASR/ML metrics such as WER, latency, F1 score, EER, and DER.
• Integrate LLMs and agentic flows into production solutions where they add value.

Skills

Machine learning / deep learning

Python

Linux

Training workflows

MLOps

Inference optimization

Real-time streaming

ASR familiarity

Tools

PyTorch

TensorFlow

Docker

gRPC

Serverless / micro-services

CrewAI

AWS Bedrock

Temporal

AgentCore

LiteLLM

Job description

Verbit, Inc. is seeking a skilled ML/AI Engineer to build, deploy, and serve ML models in production, focusing on ASR-related tasks and scalable speech solutions. The role emphasizes learning in the speech domain and improving inference performance.

The ideal candidate has 4+ years in ML, strong Python, and experience with PyTorch/TensorFlow, Docker, and cloud-based MLOps. You will work to optimize latency and integrate advanced AI features across Verbit’s platform.