Research Engineer - Speech & Audio ML Advancements

Hume AI, Inc.; New York (NY); On-site; USD 140,000 - 210,000; Full time

Job summary

Hume AI is seeking a Research Engineer to advance speech-language and audio models through hands-on development, large-scale training, and rigorous evaluation. You will build software for distributed training, inference, and benchmarking, collaborating with customer teams to test ideas at scale.

You will contribute to experiments from question to validated improvements, spanning in-house and customer models, with emphasis on reproducibility and robust evaluation.

Qualifications

• At least two years of experience contributing to model training or fine-tuning with large-scale datasets in speech, audio, or multimodal ML.
• Strong Python and PyTorch skills; able to develop and maintain training and evaluation code.
• Experience building research systems beyond notebooks or prototypes.
• Strong experimental judgment with baselines and evaluation.
• Publications or open-source contributions demonstrating research and modeling ability.
• Ability to iterate quickly on uncertain research directions and own open-ended problems.
• Clear written and verbal communication with research and customer teams.

Responsibilities

• Investigate research questions in speech-language, audio, and multimodal ML and design experiments.
• Train, fine-tune, validate and develop models in-house and for customers.
• Build reliable software for distributed training, inference, benchmarking.
• Develop data workflows for large-scale datasets.
• Design evaluation protocols and benchmarks.
• Analyze datasets and model outputs to guide next steps.
• Collaborate with customer teams on training and evaluation.
• Document experimental designs and results for reproducibility.

Job description

Hume AI is seeking a Research Engineer to advance speech-language and audio models through hands-on development, large-scale training, and rigorous evaluation. You will build software for distributed training, inference, and benchmarking, collaborating with customer teams to test ideas at scale.

You will contribute to experiments from question to validated improvements, spanning in-house and customer models, with emphasis on reproducibility and robust evaluation.