Research Intern

Who we are

At MARZ we have been working for two years to build THE best in-house lip-syncing tool to break down language barriers in e-learning, advertising, and Hollywood. LipDub at its core modifies given footage to match any supplied audio track, allowing for use cases such as localization, ADR, scaled marketing, and many more. In the market today, MARZ is the premier provider of accurately lip-synced high-fidelity video dubbing, and we’re not stopping here. LipDub algorithms are developed entirely in-house by some of the strongest researchers and leaders in computer graphics and generative modelling.

Who we need

MARZ offers the opportunity for students to work with and learn from our dynamic, growing, and passionate Technology team. Some of the responsibilities of these interns include:

• Data Labeling: Use existing tools to label specific parts of the face in videos. Accuracy and attention to detail are key as this data will be used in downstream tasks.

• Dataset Expansion: Source and download new open-source video content to expand the dataset, ensuring diversity in facial features and backgrounds.

• Data Analysis: Use basic statistical techniques to evaluate the distribution of the dataset (e.g., understanding the prevalence of certain poses or features like gender, face shapes, etc).

• Run Preprocessing Scripts: Execute Python scripts to process and prepare labelled data. Familiarity with file handling and input/output operations is a bonus.

What you bring

• Education: Pursuing a degree in Data Science, Computer Science, Mathematics, Engineering, or a related field (Year 2 or higher).

• Technical Skills: Basic Python knowledge and familiarity with data-handling libraries.

• Detail-Oriented: Able to work meticulously and ensure accuracy in labelling tasks.

• Self-motivated: Comfortable working independently and proactively sourcing new data as needed.

Bonus Skills: Experience with video processing or familiarity with tools for data labelling would be a plus.