Systems Performance Engineer, Agentic AI Workloads – New College Grad 2026

We're looking for a Deep Learning Architect to join our team working at the cutting edge of AI infrastructure. As agentic LLM workloads reshape the demands placed on modern datacenters, we need engineers who can model, simulate, and reason about complex system-level traffic at scale.

In this role, you will build and run simulations that capture the traffic dynamics of agentic AI workloads, mine the results for actionable insights, and help guide architectural decisions for next-generation datacenter and GPU systems.

Key Responsibilities:

• Develop and extend C++ and Python simulators that model system-level network and compute traffic for agentic LLM workloads in datacenter environments
• Characterize real-world LLM serving workloads and distill them into representative simulator inputs
• Run simulations at scale and apply statistical techniques to post-process and interpret results
• Identify performance bottlenecks and translate findings into concrete architectural recommendations
• Collaborate with hardware, software, and research teams to influence the design of future AI systems

Requirements:

• Pursuing or recently completed a MS, or PhD in CS, EE, Mathematics, or a related field (or equivalent experience)
• Strong programming skills in C++ and Python
• Solid foundations in queueing theory and traffic modeling (e.g., Erlang models, Little's Law)
• Strong statistics background: characterise huge datasets with percentiles, distributions, and clustering techniques such as K-means
• Understanding of deep learning fundamentals, LLMs, and modern inference serving frameworks

Nice to Have:

• Hands-on experience with traffic or network simulators, even in an academic or course project context
• Familiarity with roofline modeling and performance scaling of deep learning models at the kernel level
• Experience running large-scale simulation campaigns and building data pipelines to process and visualise results
• Prior work characterising or benchmarking ML inference workloads

If you're analytically sharp, intellectually curious, and ready to have real impact, we want to hear from you.