Research Scientist Intern - Multimodal RL for VLMs

Meta; Sunnyvale (CA); On-site; USD 40,000 - 60,000; Part time

Job summary

Meta Reality Labs Research invites outstanding Research Scientist Interns to advance posttraining of vision-language and omni models. You will develop RL methods to improve multimodal reasoning, temporal understanding, and grounded decision-making for contextual AI devices.

Internships run twelve to twenty-four weeks with flexible start dates. Applicants should have a PhD in machine learning, computer vision, natural language processing, speech and audio processing, or reinforcement learning,

Qualifications

• PhD or in progress in ML, CV, NLP, or related field.
• Demonstrated research contributions via grants, fellowships, patents, or publications.
• Hands-on experience with deep learning and multimodal models, RL posttraining.
• Experience implementing DL algorithms, training models, and designing benchmarks.

Responsibilities

• Develop RL posttraining methods for vision-language and omni models.
• Design reward models and evaluation benchmarks for multimodal reasoning.
• Collaborate across teams to advance contextual AI devices.

Skills

Python

PyTorch

Reinforcement Learning

Multimodal Models

Team Collaboration

Education

PhD in ML / CV / NLP / speech & audio / RL

Tools

PyTorch

Job description

Meta Reality Labs Research invites outstanding Research Scientist Interns to advance posttraining of vision-language and omni models. You will develop RL methods to improve multimodal reasoning, temporal understanding, and grounded decision-making for contextual AI devices.

Internships run twelve to twenty-four weeks with flexible start dates. Applicants should have a PhD in machine learning, computer vision, natural language processing, speech and audio processing, or reinforcement learning,