Alibaba DAMO Academy Opens 15 Chip and AI Research Topics for Its “Alibaba Star” Talent Program

ITHome reported on August 6 that Alibaba’s DAMO Academy today announced its “Alibaba Star” elite talent recruitment and development program for 2027 graduates. DAMO Academy has opened 15 research topics this year, covering cutting-edge fields such as AI chips, novel CPU architectures, multimodal healthcare agents, and AGI decision-making. The specific topics are as follows: Chip Technologies Design and Implementation of Programming Models for AI Chips Develop an entirely new programming language and compiler toolchain for Alibaba’s proprietary AI chips. This is not simply a matter of “porting an LLVM backend,” but involves the entire process, from DSL design, compiler frontends, and intermediate optimization frameworks to backend code generation. Design and Implementation of AI Compilers Develop graph compilers and operator compilers for proprietary AI chips, addressing technical challenges such as automatic model parallelization, distributed data processing, and the overlapping of computation and communication. Research on AI SoC Technologies for Large Models Conduct systematic research and engineering implementation of advanced architectures and key technologies for next-generation AI SoCs, targeting large-model training and inference as well as data-center-scale AI computing scenarios. The work will focus on technological upgrades and the development of core capabilities, driving breakthrough innovation in chip architecture, advanced packaging, and hardware-software system coordination. Improving AI Chip Software Development Efficiency with AI Agents Use AI agents to reshape software development, from automatic code generation and intelligent code review to automated testing, documentation generation, fault diagnosis, and intelligent orchestration of the entire development workflow. Optimization of Inference Systems for AI Chips Build high-performance inference systems for proprietary AI chips and design the core modules of inference engines, seeking optimal solutions across algorithms, systems, and hardware. The goal is to achieve industry-leading inference throughput and latency for large models with hundreds of billions of parameters. Hardware-Software Co-Design for AI Chips Bridge hardware and software by participating in the hardware-software co-design of proprietary AI chips. Responsibilities include optimizing hardware architecture based on feedback from the software stack, defining hardware-software interface specifications, developing performance models to guide architectural decisions, and conducting system-level verification before tape-out. Innovative, Energy-Efficient AI Chip Architectures for Product Integration Design and evaluate AI hardware architectures based on business development requirements and advanced semiconductor, architecture, and systems technologies, and bring chip products to market. The work includes, but is not limited to, system and chip architecture and microarchitecture design, modeling and quantitative evaluation, and algorithm and application analysis. Exploring CPU Architectures for AI Agents Participate in architectural research for next-generation, high-performance RISC-V CPUs and explore new CPU architectures for the era of agentic AI. Study the characteristics of agent workloads, identify performance bottlenecks, design new RISC-V microarchitectures, contribute extensively to RISC-V International, and participate in defining industry standards. AI-Driven Methods for High-Performance RISC-V Compiler Optimization Advance the design, development, and optimization of compiler systems based on the RISC-V instruction set. Use AI models to replace or optimize key compiler components, enabling more intelligent optimization decisions; design scenario-specific instruction set extensions; and participate in standards development at RISC-V International. AI Technologies Research on Foundation Models for Healthcare AI Research foundation-model technologies related to medical imaging AI, including but not limited to multimodal large models, foundation models, AIGC generative models, AI agents, and large-model optimization. Address clinical healthcare challenges such as pan-cancer early screening, imaging report generation, medical image data synthesis, large models for medical image analysis, and healthcare AI agents. Research on Multimodal Healthcare Agents Lead the end-to-end research and development of cardiovascular disease diagnostic agents and pathology diagnostic agents based on visual foundation models and multimodal agent technologies. Design and implement fine-tuning and adaptation solutions for visual foundation models in medical imaging; lead the architectural design and development of core modules for multimodal agent systems; and explore and implement the use of reinforcement learning in agent training. Multimodal Large Models and AI Agents for Healthcare Develop multimodal large models for healthcare scenarios, including medical vision-language models, and explore the construction of healthcare AI agents. Work with clinicians to collect and annotate data and conduct clinical validation, driving the deployment and application of models in real-world business scenarios. Healthcare AI Foundation Models and Agents Develop healthcare AI foundation models and agent systems. Design and develop next-generation intelligent healthcare systems with perception, comprehension, reasoning, and decision-making capabilities for real-world medical scenarios. Analyze business and technical requirements, design appropriate technical solutions and R&D roadmaps for real-world healthcare settings, and drive the practical adoption of healthcare AI. Research on Artificial General Intelligence Technologies for Decision-Making Research long-term memory and self-awareness modeling, advanced reasoning mechanisms, autonomous agents, continual learning, and lifelong evolution for industrial scenarios including electric power and energy, oil, petrochemicals and pipeline networks, aviation, high-speed rail, and ports. Exploration and Productization of Cutting-Edge AI Video Algorithms and Standards Jointly develop advanced AI video technologies, drive their large-scale deployment, participate in the development of the latest international video standards, and accelerate the industrialization of AI video technology.