Applied ML Engineer: Shipping Multimodal Models

Trace; United States; On-site; USD 120,000 - 170,000; Full time

Job summary

Trace Labs is building the data infrastructure for physical AI. This role sits between ML engineering and research, bringing strong ML engineering skills, a research mindset, and deep learning fundamentals to the hardest problems we see at Trace, then shipping the answers into production.

You’ll work closely with our computer vision team on multimodal data—from video to sensor streams and language—and you’ll own models end to end, from data and training to evaluation and deployment into our

Qualifications

• 2+ years of industry hands-on experience training deep learning models on real-world data.
• Strong proficiency in Python and a modern deep learning framework such as PyTorch or JAX.
• Experience with data from video, time series, or sensor streams.

Responsibilities

• Own models end to end: data, training, evaluation, and deployment into our annotation pipeline.
• Build training pipelines for large, multimodal datasets, including video, sensor streams, and language.
• Partner with our Head of Engineering and the CV team to get models into production and keep them improving.
• Start with the simplest approach that works, then make it better. Pull from recent papers when they help you ship, not as the end goal.
• Move quickly between very different problems, and build the tooling you need along the way.

Skills

Python

Deep learning

ML model development

Research mindset

Production deployment

Data handling

Education

BS in Computer Science, Electrical Engineering, Mathematics, Aerospace, or related field

MS is a plus

Tools

PyTorch

JAX

Job description

Trace Labs is building the data infrastructure for physical AI. This role sits between ML engineering and research, bringing strong ML engineering skills, a research mindset, and deep learning fundamentals to the hardest problems we see at Trace, then shipping the answers into production.

You’ll work closely with our computer vision team on multimodal data—from video to sensor streams and language—and you’ll own models end to end, from data and training to evaluation and deployment into our