Meta’s Elite Recommendations Lab

Meta reorganized its Recommendation Systems org last fall and quietly spun up a specialized research unit called MRS Research in October 2025 to pursue long‑term AI advances for Facebook and Instagram feeds. (letsdatascience.com) The unit is led by Yang Song, who joined Meta in November 2025 after running user‑growth and recommendation efforts at TikTok and is listed internally as vice president of recommendation research. (africa.businessinsider.com) Reporting and hiring traces show MRS Research has been hiring senior engineers and scientists from TikTok, Amazon, OpenAI and Google—public posts and LinkedIn moves name hires from Amazon AI teams and researchers with OpenAI/Google backgrounds. (letsdatascience.com) Active Meta job descriptions tied to MRS and MRS AI explicitly list work on retrieval, candidate generation, ranking, large language models, generative methods, and monetization tasks such as click‑through‑rate prediction and ad‑ranking. (simplyhired.com) MRS Research sits inside Meta’s Modern Recommendation Systems that supply home‑feed ranking across Facebook, Instagram and Threads and works closely with Ads teams—recommendation improvements feed into the ad pipeline that helped generate $200.97B in revenue for Meta in 2025. (jobs.anitab.org) Public job specs and the Business Insider report make the recruiting signal explicit: expect interviews for these roles to probe large‑scale retrieval/ranking architectures, offline metrics (e.g., NDCG/CTR), and production challenges around model latency and monetization—skills being listed on current openings. (businessinsider.com)