{"slug": "ai-native-computing-systems-from-computing-for-ai-to-ai-for-computing", "title": "AI-Native Computing Systems: From Computing for AI to AI for Computing", "summary": "Zishen Wan, an incoming assistant professor at Columbia CS (Fall 2026) and currently a postdoc at Harvard, will present his research on AI-native computing systems, covering workload characterization and cross-layer co-design for physical and neuro-symbolic AI, as well as using AI agents to design computer architectures. His work, recognized with awards including the ACM SIGDA Outstanding PhD Dissertation Award and Best Paper Awards at DAC and CAL, has been adopted by Intel, IBM, and Google.", "body_md": "# AI-Native Computing Systems: From Computing for AI to AI for Computing\n\n## Abstract\n\nAI is rapidly evolving from individual models into dynamic, heterogeneous, and autonomous systems, creating new challenges for computer architecture and systems. Embodied and agentic AI increasingly combine models, tools, memory, planning, and verification into complex workflows, demanding new approaches beyond optimizing individual models and kernels. At the same time, advances in AI reasoning create a complementary opportunity: AI itself can become part of the architecture and system design process.\n\nIn this talk, I will present our work toward AI-native computer architectures and systems from these two directions. I will first discuss workload characterization and cross-layer co-design for physical and neuro-symbolic AI, spanning runtimes, hardware architectures, and silicon tapeouts. I will then show how AI agents can reason about, evaluate, and design computer architectures, moving from architectural benchmarking and design-space exploration toward architecture discovery beyond existing simulator capabilities.\n\nTogether, these efforts point toward a future in which AI is both the workload and the designer, enabling computing systems that can adapt to emerging forms of intelligence and increasingly participate in their own design and optimization.\n\n## Bio\n\nZishen Wan is an incoming assistant professor at Columbia CS (Fall 2026). He is currently a postdoc at Harvard working with Prof. Vijay Janapa Reddi, and received his PhD from Georgia Tech ECE advised by Profs. Tushar Krishna and Arijit Raychowdhury. His research focuses on computer architecture, with an emphasis on cross-stack co-design of systems, architectures, and silicon for embodied intelligence, as well as agentic AI for computing system design. His work appears in venues including ASPLOS, MICRO, HPCA, JSSC, ISSCC, and DAC, and has been recognized with ACM SIGDA Outstanding PhD Dissertation Award, Best Paper Awards at DAC and CAL, IEEE Micro Top Picks, First Place in DAC PhD Forum, Qualcomm and Baidu PhD Fellowships, WAIC Yunfan Award, and ML and Systems Rising Star. His research has been adopted by industry partners including Intel, IBM, and Google. For more information, please visit [https://zishenwan.github.io/](https://zishenwan.github.io/).", "url": "https://wpnews.pro/news/ai-native-computing-systems-from-computing-for-ai-to-ai-for-computing", "canonical_source": "https://systems.seas.harvard.edu/seminar/2026-08-10-zishen-wan/", "published_at": "2026-08-10 14:00:00+00:00", "updated_at": "2026-08-10 14:36:02.210452+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "ai-agents", "computer-vision"], "entities": ["Zishen Wan", "Columbia CS", "Harvard", "Georgia Tech", "ACM SIGDA", "Intel", "IBM", "Google"], "alternates": {"html": "https://wpnews.pro/news/ai-native-computing-systems-from-computing-for-ai-to-ai-for-computing", "markdown": "https://wpnews.pro/news/ai-native-computing-systems-from-computing-for-ai-to-ai-for-computing.md", "text": "https://wpnews.pro/news/ai-native-computing-systems-from-computing-for-ai-to-ai-for-computing.txt", "jsonld": "https://wpnews.pro/news/ai-native-computing-systems-from-computing-for-ai-to-ai-for-computing.jsonld"}}