Machine Learning Engineer - Speech & Natural Language

Role overview A machine learning engineer with deep speech and language expertise is needed to own the natural language stack running on fielded edge hardware with no cloud connection. The role covers the full path from architecture selection through training, optimization, and shipping a hardened inference runtime that fits inside a constrained compute budget. It is a build role, not a foundational research role, and the bar is demonstrated production results rather than years of experience.

Responsibilities - Assess candidate architectures and modeling approaches against operational criteria such as accuracy under degraded audio, latency, and footprint, and defend the choice with data. - Build and maintain the synthetic data generation tooling, augmentation pipelines, and field-data curation that drive model accuracy. - Train and fine-tune speech and language models that handle noisy environments, degraded audio, and domain-specific operator vocabulary. - Reduce latency and memory through principled optimization, proving each gain with benchmarks. - Convert research-grade code into a deployable production runtime with experimental scaffolding stripped out. - Define how success is measured, build evaluation that reflects whether an operator's command actually worked, and turn field failures into test cases, training data, and durable fixes.

Requirements - Substantial hands-on experience training, optimizing, and deploying ML models in production systems, with a track record of results rather than tenure. - Deep, hands-on experience across both speech recognition and natural language understanding, including domain adaptation. - Deep proficiency in Python and modern deep learning frameworks, including custom training loops, data pipelines, and evaluation harnesses. - Demonstrated experience making models meaningfully faster or smaller, with the benchmarks to back it up. - Hands-on work moving models out of framework and research code into an efficient, deployable inference path. - Degree in computer science, electrical engineering, computational linguistics, or a related technical field, or equivalent proof of capability. - U.S. citizenship and current U.S. residency are required.

Nice to have - Experience deploying models on embedded or resource-constrained GPU hardware. - Background in defense, robotics, or autonomy. - Familiarity with model lifecycle tooling such as experiment tracking, model registries, and automated evaluation.