Multimodal AI Research Scientist for Video Editing

Descript; San Francisco (CA); Hybrid; USD 197,000 - 263,000; Full time

Benefits offered by this job

Equity

Job summary

Descript's Research team builds models behind the product's standout features, including Video Regenerate and lipsync. This role focuses on multimodal understanding, training models to perceive edited media the way a human editor does, enabling better evaluation and smarter editing decisions.

The candidate will design and implement deep learning solutions, evaluate experiments, and help ship production features.

Qualifications

• Proven ability to design and implement deep learning algorithms.
• Strong programming skills and fluency in PyTorch.
• Track record of generating new ideas and shipping results.

Responsibilities

• Train models from scratch or fine-tune existing foundation models.
• Design benchmarks and evaluation methods for edited media.
• Ship production features in collaboration with engineering teams.
• Mentor or lead researchers as needed for senior roles.

Skills

Deep learning

PyTorch

Experimental judgment

Communication

Publications

Education

PhD or Master's in deep learning or related field

Job description

Descript's Research team builds models behind the product's standout features, including Video Regenerate and lipsync. This role focuses on multimodal understanding, training models to perceive edited media the way a human editor does, enabling better evaluation and smarter editing decisions.

The candidate will design and implement deep learning solutions, evaluate experiments, and help ship production features.