Cambridge Residency Programme – Researcher in Agentic AI Systems & Infrastructure  

Conduct original research on the design, architecture, and optimization of agentic AI systems, focusing on memory, communication, and orchestration. Prototype new components for multiagent inference with system-level optimizations (e.g. shared latent memory/KV-cache, agent-level parallelism) using relevant framework tools and inference backends like vLLM and SGLang. Explore ML & systems codesign opportunities, such as aligning model capabilities with systems constraints, hardware characteristics, and orchestration strategies, and using Pytorch and other relevant tools of LLM fine-tuning on GPU clusters. Evaluate proposed ideas through real-system experiments, large-scale benchmark evaluation, and empirical studies on real workloads. Work closely with a multidisciplinary team to address both fundamental and applied research challenges. Communicate results clearly, sharing insights with the wider team and partner groups Contribute to an open, multidisciplinary research environment PhD (or near completion) in Computer Science, Machine Learning, Electrical Engineering, or a related field Strong background in ML-systems co-design, AI inference systems, or machine learning systems. Demonstrated ability to conduct independent, highimpact research, evidenced by publications, opensource systems, or deployed artifacts. Ability to work effectively in collaborative, crossdisciplinary research teams. Familiarity with modern agentic systems, orchestration patterns, or largescale ML infrastructure. Experience in model post-training, reinforcement learning / evolution strategies, or supervised fine-tuning. Experience in building high-performance LLM inference systems using SGLang or vLLM.