Research Scientist, Contextual AI and Multimodal Agents

Meta•Redmond, WA

1d•$219,000 - $301,000

About The Position

Meta is seeking a Research Scientist to advance foundational machine learning research that powers products used by billions of people worldwide. In this role, you will identify and solve the hardest open problems in core ML — spanning areas such as multimodal learning, foundation models, large-scale model training, efficient inference, and agentic systems — and translate breakthrough research into systems that fundamentally improve Meta's products and platforms. This is a senior individual-contributor role for a recognized technical authority. You will define the long-term technical vision for your research area, serve as a principal driver of Meta's most ambitious bets in machine learning, and align work across disciplines and organizations around a coherent multi-year direction. You will remain technically deep—prototyping critical-path ideas, stress-testing results, and making consequential technical decisions—while multiplying the impact of other researchers and engineers through mentorship, collaboration, and clear scientific judgment.

Requirements

Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience
PhD in Computer Science, Artificial Intelligence, Machine Learning, Computer Engineering, Electrical Engineering, or a related technical field
12+ years of research or applied-development experience in machine learning, artificial intelligence, or related areas
Recognized authority in a relevant research area, with demonstrated breadth across adjacent areas such as multimodal learning, foundation models, efficient AI systems, or agentic AI
Experience setting the technical direction for a research area and leading multiple complex initiatives whose methods, systems, or strategy were adopted beyond the immediate team
Hands-on experience building and evaluating foundation-model systems, such as large language models or multimodal models using techniques like long-context reasoning, retrieval, or tool use
Demonstrated success carrying research from inception into shipped product capabilities, widely adopted platforms, commercialized technology, or open-source systems with a meaningful user community
Sustained record of influential research contributions demonstrated through publications at leading venues, widely adopted research artifacts, patents, or comparable impact on the field
Experience building and leading durable cross-organizational collaborations, including partnerships with external research institutions or industry, and mentoring senior researchers or engineers beyond the immediate team
Experience with Python and modern machine-learning frameworks such as PyTorch or JAX

Nice To Haves

Reinforcement learning for reasoning, agent self-improvement, or adaptive behavior
Experience driving ML efficiency and scalability improvements across large distributed training or inference systems, including model compression, quantization, distillation, and adaptive computation
Research contributions that have introduced new paradigms or techniques subsequently adopted broadly within the ML community, such as novel architectures, training methods, or theoretical frameworks
Implementation or deployment of ML systems on mobile, embedded, or other resource-constrained platforms
Founding or growing research communities, open-source ecosystems, or strategic industry-academic collaborations
Experience collaborating with privacy, security, or integrity teams to design ML systems that are robust, safe, and compliant with regulatory requirements
Design of datasets, benchmarks, or evaluation methods for foundation models, multimodal understanding, or agent capability

Responsibilities

Define and drive a multi-year research strategy for core machine learning, connecting areas such as multimodal understanding, foundation models, efficient training and inference, and agentic reasoning into a coherent research portfolio
Serve as a principal technical driver for critical, high-risk research bets with multi-year potential, carrying them from problem formulation through scientific validation, system demonstration, and productization
Lead ambitious research initiatives across algorithm and system design, implementation, experimentation, evaluation, and transfer into reusable platforms or product capabilities
Develop novel ML algorithms and architectures—including training, post-training, reinforcement-learning, and evaluation methods—that advance the state of the art in areas such as large-scale optimization, representation learning, and generalization
Advance efficient training and inference through model compression, quantization, distillation, adaptive computation, and hardware-software co-design for on-device and large-scale deployment
Establish shared datasets, benchmarks, evaluation frameworks, and research platforms that raise the quality bar across multiple teams and product areas
Identify new product opportunities and build durable collaborations across Meta, academia, and industry to move research into widely useful technology
Mentor researchers and engineers across teams, provide technical counsel on consequential decisions, and raise the scientific and engineering bar for the broader organization
Publish influential research, contribute open-source software and datasets where appropriate, and strengthen Meta's position in the research community through talks, collaborations, and service
Partner with legal, policy, and compliance teams to ensure that ML research and deployment practices uphold privacy, security, and responsible AI standards