Research Scientist Intern, AI for Egocentric Representation Learning (PhD)

Meta - Burlingame, CA

Full-time - Intern

Burlingame, CA

Web Search Portals, Libraries, Archives, and Other Information Services

About the position

Reality Labs Research is seeking an intern to assist in developing next-generation assistance systems for contextual and adaptive AR/VR environments. The role focuses on multimodal learning and self-supervised learning, with an emphasis on AI for Egocentric Representation Learning. Interns will collaborate with researchers to tackle significant problems and contribute to publishable research in top-tier conferences.

Responsibilities

• Develop, implement, and evaluate methods for learning robust representations from multi-modal egocentric data (e.g., video, audio, inertial measurement units).
• Utilize Meta's large infrastructure to scale and expedite experimentation.
• Write modular research code for reuse in various contexts.
• Collaborate with other researchers on projects.
• Deliver clear, compelling, and creative solutions to complex problems at scale.
• Aim for research outcomes suitable for publication in top-tier ML or CV conferences.

Requirements

• Currently pursuing or has a PhD in machine learning, computer vision, speech processing, applied statistics, computational neuroscience, or a relevant technical field.
• Excellent research skills in problem definition, solution exploration, and result analysis and presentation.
• Proficiency in Python and machine learning libraries (numpy, scikit-learn, scipy, pandas, matplotlib, tensorflow, pytorch).
• Deep understanding of transfer learning, few-shot learning, zero-shot learning, continual learning, self-supervised learning, or multi- or cross-modal learning.
• Strong interpersonal skills for cross-group and cross-culture collaboration.
• Must obtain and maintain work authorization in the country of employment.

Nice-to-haves

• Proven track record of significant achievements (grants, fellowships, patents, first-authored publications at leading conferences).
• Experience with deep metric learning or neural net embedding methods.
• Experience with vision-based input recognition systems (e.g., hand tracking, body pose estimation).
• Experience working with time-sequence sensor data (e.g., IMU and audio).
• Experience in cross-functional team environments.
• Intent to return to degree program after the internship.
• Availability for a minimum of 16 consecutive weeks.

Benefits

• Competitive monthly compensation ranging from $7,500 to $11,333.
• Benefits package including health insurance and other perks.