United States Role overview
About this role
Join the pioneering AI Foundations Research team at IBM Research and contribute to shaping the future foundations of artificial intelligence. Our group of scientists, engineers, and designers is dedicated to conducting end-to-end research that delivers real-world AI impact through a rigorous, responsible, and open innovation framework. We are the team behind IBM's Granite open-source models, and our work spans multi-modal (vision, speech, language, code) and multi-lingual systems, data quality, data generation, novel model architectures, training recipes, governance and trustworthiness, and programming models for interacting with models. As an intern, you will explore cutting-edge research areas, including new algorithms for training, fine-tuning, reinforcement learning, and inference time scaling, alongside pioneering work in generative computing and generative programming for language, code, and other modalities, all within a collaborative environment that bridges fundamental science and transformative engineering. As an AI Research Scientist Intern, you will engage in the full research lifecycle to pioneer new advancements in artificial intelligence. Your role will involve identifying core challenges, designing novel prototype solutions, and validating them through rigorous experimentation. You will conduct foundational research in critical areas such as novel training algorithms, efficient fine-tuning, reinforcement learning, and inference-time scaling for multi-modal models. Collaborating in small, mentored teams, you will be responsible for shepherding projects from ideation to completion. A key objective is to disseminate significant results through publications in leading conferences and patent applications, contributing to both the academic community and IBM's open innovation initiatives. Research Experience with Modern ML: Hands-on experience with and a deep theoretical understanding of Large Language Models (LLMs), Vision-Language Models (VLMs), and Transformer-based architectures. Strong Programming & Prototyping Skills: Proficiency in Python for rapid prototyping, experimental setup, and implementing large-scale machine learning systems. Rigorous Analytical & Problem-Solving Abilities: Demonstrated strength in quantitative analysis, designing experiments, and deconstructing complex research problems. Proven Research & Communication Skills: A track record of innovation (evidenced by peer-reviewed publications or strong preprints) and the ability to clearly articulate complex technical concepts in both writing and presentations. Collaborative Research Mindset: A team-oriented approach with a commitment to rigorous, reproducible, and well-documented research practices. Research Excellence: Strong publication record in top-tier conferences (e.g., NeurIPS, ICML, ICLR, CVPR, AAAI). Technical Proficiency: Advanced expertise in ML frameworks (PyTorch) and full-cycle development of algorithms and systems. Specialized Skills: Hands-on experience with generative AI (LLMs) and multimodal models, from training to testing. Communication: Demonstrated ability to present complex research and build high-impact technical demonstrations.