AI Research Scientist, Post-Training - Meta Superintelligence Labs at Meta

Menlo Park, California, United States -

Full Time

Start Date

Immediate

Expiry Date

29 May, 26

Salary

217000.0

Posted On

28 Feb, 26

Experience

2 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

Research Science, Post-Training, Data Curation, SFT, RLHF, Model Quality, Scientific Judgment, Data Strategy, Agentic Tasks, Deep Research, Coding, Deep Learning, NLP, DPO, Alignment

Industry

Software Development

Description

Meta is seeking Research Scientists to join the Post-Training team within Meta Superintelligence Labs (MSL). High-quality data is the core of AI progress at MSL, fueling the complex capabilities we build, how our models reason, and how they interact with the world. As a Research Scientist, you will provide the technical vision to design, generate, and curate the critical post-training data (SFT, RLHF) that aligns and enhances our frontier AI systems. You will conduct research to develop and optimize post-training recipes that directly improve model quality. This is a technical research role requiring sound scientific judgment, creativity, and the ability to drive ambitious research agendas with independence. The data strategies you develop will directly influence research direction and major model lines within MSL, making data quality, methodological rigor, and clear communication important. You will collaborate closely with technical leadership to ensure our data pipelines capture the most important capabilities—ranging from expert domains (STEM, GDP-valuable tasks, finance, legal, health) to advanced agentic tasks (search, Deep research, computer use, coding, UI generation, and shopping agents). We are looking for exceptional research talent—researchers who have shaped the field of machine learning and are ready to do so again at the frontier of AI. If you are passionate about defining how we teach and align AI systems and want to shape the scientific foundations of frontier AI development, we encourage you to apply for this exciting opportunity at the core of MSL. Responsibilities Design novel methodologies for post-training data collection, curation, and synthetic data generation Define data quality frameworks and alignment strategies that guide capability development across MSL, particularly for complex reasoning and agentic behaviors Drive the scientific vision for eliciting high-quality data in expert domains (finance, legal, health, STEM) and complex agentic trajectories (Deep research, computer use, UI generation) Conduct research to develop and optimize post-training recipes that directly improve model quality Partner with cross-functional research teams across product and model training to identify and prioritize gaps in model capabilities Contribute to research workstreams that shape the long-term direction of data-centric AI at MSL, working independently while also contributing to team goals and organizational priorities Minimum Qualifications Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience Ph.D. in Computer Science, Machine Learning, or a related technical field 3+ years of experience in machine learning research, with a focus on deep learning, data alignment, NLP, or related areas Demonstrated ability to lead technical research projects from conception to production Effective communication skills and experience collaborating with technical leadership Preferred Qualifications Multiple first-author publications at top-tier peer-reviewed venues (NeurIPS, ICML, ICLR, ACL, EMNLP, or similar) related to language model alignment, synthetic data generation, RLHF, or deep learning Recognized expertise in data-centric AI, post-training methodologies, or complex reasoning data Track record of research that has substantially influenced the field of deep learning Hands-on experience with language model post-training, RLHF, DPO, or related alignment techniques $154,000/year to $217,000/year + bonus + equity + benefits

Responsibilities

Research Scientists will design novel methodologies for post-training data collection, curation, and synthetic data generation, defining data quality frameworks to guide capability development across frontier AI systems. They will drive the scientific vision for eliciting high-quality data in expert domains and complex agentic trajectories while optimizing post-training recipes to improve model quality.