Meta Technology/Software

Research Scientist, Applied AI - AI Specialist

New York City, NY +4 In person Full-time

Meta is seeking applied research scientists to join the Applied AI organization (AAI) to help deliver on one of the most ambitious goals in technology: personal superintelligence for everyone. As an applied research scientist, you will apply your deep domain expertise, engineering skills, and sound judgment to make our frontier LLMs industry-leading on real-world tasks. You will identify gaps in the current model capabilities and build the complex systems, agentive pipelines, reinforcement learning environments, and evaluations that generate high-quality training signals. Working closely with Meta Superintelligence Labs, AAI contributes to advancing Meta's frontier models and AI products.

We deliberately operate like an early-stage company: a flat, AI-native organization with light processes, wide scope, and fast decisions. We are looking for applied research scientists with deep technical foundations in machine learning, systems design, or a related area and AI fluency who are comfortable with ambiguity and energized by advancing Meta's frontier AI capabilities.

Responsibilities

  • Define and lead a research initiative addressing a significant limitation in model learning, reasoning, agent behavior, or efficiency. Formulate the scientific agenda, set intermediate milestones, and make direct technical contributions to critical methodological or implementation problems
  • Establish rigorous evaluation across the initiative, including reproducible baselines, ablations, held-out tasks, realistic failure analysis, and quality-versus-cost comparisons. Ensure conclusions survive independent checks and do not rely on evaluation leakage or reward exploitation
  • Guide the design and delivery of reusable training, inference, or agent systems with partner teams. Align architectural and modeling choices, measure capability and efficiency outcomes, and support adoption with clear agreements on partners' integration and operational responsibilities
  • Raise the team's research and engineering quality through reviews, mentorship, and clear technical direction. Develop colleagues' independent ownership, help prioritize high-value work, and disseminate findings through shared software, technical reports, and publications where appropriate

Minimum qualifications

  • Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience
  • PhD in Computer Science, Artificial Intelligence, a related technical field, or equivalent applied research experience
  • 8+ years of programming experience in Python, C++, or another language relevant to scientific computing
  • Experience developing and implementing machine learning methods or AI systems, with depth in a relevant area such as model learning, reasoning, agent behavior, or model efficiency
  • Experience leading a major research or technical initiative and making direct contributions to critical methodological, modeling, or implementation problems
  • Experience in establishing reproducible models or agent evaluations using appropriate baselines, ablations, held-out tests, failure analysis, and quality-versus-cost comparisons
  • Experience designing reusable training, inference, evaluation, or agent components and evaluating architectural, performance, scalability, and reliability trade-offs
  • Experience aligning work with research, engineering, or product partners, influencing technical direction, mentoring colleagues, and communicating decisions through technical reports or design documents
  • A track record of peer-reviewed research publications with experience presenting scientific methods and findings to a broad audience

Preferred qualifications

  • Experience with LLM post-training, reinforcement learning, reward modeling, or methods for learning from environment feedback
  • Experience with agent memory, planning, tool use, or automated research and evaluation workflows for long-horizon tasks
  • Experience with model-efficiency techniques such as quantization, distillation, sparsity, or hardware-aware training and inference

Free tool / Resume tailor

Tailor your resume for this job

Paste your resume and get 3 to 8 of your own lines rewritten in the words this posting uses, with the reason for each change. The job is already filled in, and you don't need an account.

Job description provided by Meta.
View the original posting