Meta has assembled a secret team called MRS Research to overhaul its advertising algorithms, enhancing content and ad placement across its platforms, including Facebook, Instagram, and Threads, according to Business Insider.
The newly-formed team, which began operations in October 2025, is part of Meta’s Recommendation Systems division and plays a critical role in optimizing the algorithms that underpin nearly all of Meta’s $201 billion revenue in 2025.
MRS Research is led by Yang Song, who was appointed vice president of recommendation research in November 2025 after his previous tenure at TikTok. At TikTok, Song managed over 400 machine learning engineers and researchers, focusing on recommendation and user growth algorithms.
Song described the mission of MRS Research as a transformation of recommendation systems through the use of large language models, emphasizing knowledge integration and advanced reasoning.
The team has made significant hires from rival tech firms, including former Amazon AI researcher Lihong Li, OpenAI researcher Xiaolong Wang, and Google researcher Fei Sha. These recruits strengthen MRS Research’s capabilities in applied AI.
Meta clarified that MRS Research operates independently from Meta Superintelligence Labs, which focuses on broader AI model development. MRS Research is specifically committed to improving content understanding, user understanding, retrieval, and ranking related to ad placements.
The recent formation of this team coincides with Meta’s increased focus on AI-driven ad optimization. In late 2025, the company introduced a new model aimed at enhancing ad effectiveness, complemented by an enhanced ad-serving process for Instagram leveraging an Adaptive Ranking Model.
According to Morgan Stanley, Meta’s advertising revenue is projected to grow by 28% in 2026, highlighting the potential impact of these initiatives on the company’s future financial performance.
