In this role, you will conduct foundational and applied research on language understanding, generation, reasoning, and alignment, contributing directly to the development of next-generation language models that power Meta's AI products and services at global scale. You will drive high-impact research from conception through publication and production integration, collaborating with world-class researchers and engineers to shape the trajectory of language AI.
Responsibilities
Design and execute original research on large language models, covering areas such as post-training, instruction tuning, reasoning, alignment, agentic tool calling and evaluation
• Develop novel architectures, training objectives, and data curation strategies to improve language model capabilities and reliability
• Lead end-to-end research projects including problem formulation, experimental design, analysis, and communication of findings to technical and non-technical stakeholders
• Build and maintain rigorous evaluation frameworks and benchmarks to measure language model performance across diverse tasks and domains
• Collaborate with cross-functional partners in engineering, product, and policy to translate research advances into production language AI systems
• Identify and drive improvements to training pipelines, data quality, and model efficiency using instrumentation, profiling, and systematic experimentation
• Mentor other researchers and engineers on the team, providing technical guidance on language model research methodology and best practices
• Leverage AI-assisted tools and workflows to accelerate research iteration, code quality, and cross-disciplinary collaboration
Minimum Qualifications
• Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience
• 8+ years of research or industry experience in natural language processing, large language models, or a closely related area of machine learning
• Experience leading complex, multi-stage research projects from problem definition through publication or production deployment
• Track record of publications at peer-reviewed NLP or machine learning conferences such as ACL, EMNLP, NAACL, NeurIPS, ICML, or ICLR
• Hands-on experience training, fine-tuning, or evaluating large language models using frameworks such as PyTorch
• Experience communicating research findings and technical decisions in writing to both research and engineering audiences, including design documents and technical reports
Preferred Qualifications
• Experience bridging foundational language model research with production systems, including model evaluation at scale and staged rollout using feature flagging and A/B testing frameworks
• Experience designing and curating large-scale pretraining or fine-tuning datasets, including data quality filtering, deduplication, and domain mixing strategies
• Experience with large-scale distributed training of language models across hundreds or thousands of GPUs, including optimization of training stability and throughput
• Demonstrated research contributions in areas such as reasoning, instruction following, reinforcement learning from human feedback, or model alignment and safety
