Staff Machine Learning Engineer

Otter.ai•Mountain View, CA

5h•$210,000 - $275,000•Hybrid

About The Position

Do you want to lead projects to build and deploy cutting-edge AI technology to help people get unparalleled value from meetings and conversations? Join our core AI team responsible for ML and work alongside industry-veteran scientists and engineers. As a Staff Machine Learning Engineer, you’ll bring your strong software engineering mindset to machine learning in order to scale and optimize our ML systems—creating and transforming innovative research into production-ready features that power Otter’s summarization and conversational intelligence products.

Requirements

Holds a Bachelor’s or Master’s degree in Computer Science or a related field with 10+ years of relevant industry experience; PhD is preferred.
Has deep, hands-on experience building and fine-tuning large language or foundation models, with production experience in ASR, TTS, multimodal, or modern LLM/NLP systems, and a strong understanding of model failure modes and trade-offs.
Demonstrates strong command of modern ML research, with the ability to critically evaluate new papers and drive innovation by identifying what is production-worthy versus experimental.
Has extensive experience deploying, scaling, monitoring, and operating ML systems in production across training, inference, and serving infrastructure, including model versioning, rollback strategies, and performance regression detection while balancing cost, latency, and reliability constraints.
Is comfortable working with large-scale speech and conversational datasets, including data preprocessing, augmentation, quality analysis, and labeling strategies to support model training and evaluation.
Is highly effective at cross-functional collaboration, working end-to-end with product, infra, research, and data teams to deliver outcomes, not just models.
Can lead technical projects independently, driving clarity in ambiguous problem spaces and making sound architectural decisions.
Has experience with or strong interest in agentic systems, tool-use frameworks, or multi-model orchestration.
Has demonstrated ability to influence beyond their immediate team, shaping technical direction, standards, or long-term strategy.

Nice To Haves

Experience with personalization, recommendation systems, or user modeling is a plus

Responsibilities

Architect, build, and evolve large-scale SID / ASR / NLP / LLM systems that power mission-critical product experiences including summarization, chat, and speech understanding across millions of conversations.
Lead the design and implementation of training, fine-tuning, post-training, and inference strategies for large language and speech models using PyTorch and/or JAX, making principled trade-offs across quality, latency, cost, and reliability.
Design and improve model architectures, loss functions, decoding strategies, and training techniques for speech and language models, informed by both research and production constraints.
Own end-to-end ML system lifecycles, from research prototyping through production deployment, monitoring, iteration, and long-term maintenance.
Partner deeply with product, and infrastructure teams to develop and translate cutting-edge research into scalable, production-grade systems that deliver measurable user and business impact.
Drive system-level improvements in model performance, robustness, observability, and operational excellence using real-world conversational data at scale.
Set technical direction and best practices for ML infrastructure, data pipelines, evaluation frameworks, and deployment workflows in a cloud environment.
Identify and resolve complex, ambiguous problems in model behavior, data quality, scaling, and system interactions, often before they surface as user-visible issues.
Mentor and elevate other engineers, influencing team standards, reviewing designs, and contributing to a culture of strong technical decision-making and execution.
Influence applied research and technical roadmaps by identifying promising speech and multimodal modeling approaches, and driving their validation and adoption into production systems