Applied Research Scientist AI Research
Descript is an active AI video and audio editing company whose editor lets users create, edit, and share media through text-based and AI-assisted workflows.
Funding history
About Descript
Descript provides an AI video and audio editor: users record or upload media, receive a transcript, and edit the underlying media by editing text. Its current product also includes an AI co-editor (Underlord), recording, collaboration, media enhancement, and publishing tools.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will build vision-language systems that help editing features understand visual and audio project content. You will design quality evaluations, create datasets including synthetic data, train or fine-tune specialized models, and take prototypes into production. You will also identify research directions and may publish work at academic venues.
Requirements
- Ability to design and implement deep learning algorithms
- Strong programming skills and deep PyTorch fluency
- Track record of generating machine-learning ideas
- Strong experimental judgment
- Clear written and verbal communication
- PhD or Master's in deep learning or a related field, or equivalent experience
- Lead or first-author publication at a top research venue or key role shipping a deep-learning production feature
Responsibilities
- Build vision-language systems for visual and audio understanding
- Design benchmarks and evaluations for editorial quality, cost, and latency
- Build datasets and generate synthetic data
- Train specialized models or fine-tune foundation models
- Take models from prototype to production
- Identify research directions for product features
- Publish research work at academic venues if desired
Benefits
- Equity
- Generous healthcare package
- 401k matching program
- Catered lunches
- Flexible vacation time
- Opportunities for in-person collaboration for remote employees