AI Researcher (Infrastructure) III (Full Time)- United States ENG/CPO/AISWP ETR

The application window is expected to close on: 11/24/2026 Applications are accepted until further notice. Meet the Team At Foundation AI, we are leading frontier AI research across Cisco. Our mission is to advance the state of AI. We translate that research into purpose-built models, agentic systems, new algorithms, optimization methods, and scalable data processing techniques that solve real-world problems at scale. We explore new model architectures, pre-training and post-training methods, inference optimization, evaluation techniques, and data pipelines. Together, we are advancing AI at Cisco and for the broader AI community! Our Infrastructure & Systems team builds the foundation that makes this research possible. We work closely with researchers to turn evolving experimental workflows into reliable, reusable systems for data preparation, model training, post-training, evaluation, and inference. Our work spans distributed GPU computing, cloud infrastructure, research tooling, and the systems that connect experiments, datasets, models, and results. Your Impact As an AI Researcher on the infrastructure team of Foundation AI, you will design, build, and operate the platforms that help our researchers move from an idea to a reproducible experiment and a deployable model. You will make complex AI workloads easier to launch, observe, debug, and scale. You will own projects from identifying problems to building and operating solutions. This could be supporting a new training stack, improving GPU scheduling, building evaluation and rollout services, or making experiments easier to track and reproduce. Your work will help researchers test ideas faster, run experiments more reliably, and ensure we are using compute as efficiently as possible. You’ll also help enable automated research workflows, where agents build on previous results to run new experiments and conceive and test novel ideas. We value engineers who understand researcher workflows, make thoughtful architectural tradeoffs, and turn evolving requirements into simple, reusable tools. You will look for signal to guide improvements in performance and reliability, debug problems across the software and infrastructure stack, and carry solutions through adoption and continued operation. You will also have opportunities to contribute to open-source projects, technical writing, and collaborative research. In this role, you will: • Build shared platforms, services, and developer tools for pre-training, fine-tuning, reinforcement learning, evaluation, and inference. • Develop and operate GPU and CPU computing infrastructure across clouds and regions, including agent sandboxes, workload orchestration, scheduling, resource sharing, storage, and networking. • Create reproducible training and serving environments through automated container builds, dependency management, compatibility validation, and release workflows. • Build reliable rollout collection and isolated execution environments for agentic workloads, including progress tracking, failure recovery, resource cleanup, and durable outputs. • Connect experiment configurations, code, datasets, checkpoints, and evaluation results so researchers and agents can reliably reproduce runs, compare outcomes, and build on prior work. • Improve observability and performance across the stack, diagnosing bottlenecks in GPU utilization, data loading, checkpointing, distributed communication, and inference. • Partner with research and data teams to build dependable data and evaluation workflows. • Work directly with researchers to understand their workflows, debug failures, and deliver well-documented capabilities that teams adopt and use. Minimum Qualifications • Bachelors + 5 years of related experience, or Masters + 3 years of related experience, or PhD + 0 years of related experience in Computer Science, Artificial Intelligence, Machine Learning, or a related field. • Strong experience with Python, PyTorch, production-grade software engineering practices, and common AI/ML libraries • Experience with large language models, model training, fine-tuning, evaluation, inference, and optimization • Strong understanding of data pipelines, model experimentation, and scalable AI system design • Strong communication skills, with the ability to explain complex research and engineering concepts to cross-functional partners Preferred Qualifications • Experience with distributed GPU training or post-training systems, including parallelism strategies, checkpointing, recovery, and performance profiling. • Familiarity with frameworks such as Megatron, NeMo, VERL, or comparable distributed training and reinforcement learning systems. • Experience deploying and optimizing LLM serving systems such as vLLM or SGLang, including throughput, latency, memory use, and concurrency. • Experience building research platforms, experiment tracking, dataset or model catalogs, evaluation harnesses, or execution environments for AI agents. • Experience with infrastructure as code, cloud identity and access management, secure workload isolation, and operational monitoring across platforms such as AWS, GCP, or specialized GPU clouds. • A record of improving developer productivity, reliability, or compute efficiency through systems adopted by other engineers or researchers. • Contributions to open-source infrastructure or ML systems, technical publications, or familiarity with cybersecurity and security-focused AI applications. Why Cisco? At Cisco, we’re revolutionizing how data and infrastructure connect and protect organizations in the AI era – and beyond. We’ve been innovating fearlessly for 40 years to create solutions that power how humans and technology work together across the physical and digital worlds. These solutions provide customers with unparalleled security, visibility, and insights across the entire digital footprint. Fueled by the depth and breadth of our technology, we experiment and create meaningful s