The AI Native Hardware & Software Codesign internship 2027 places interns within the AI Native Systems research team to explore hardware and software co-design methodologies. Participants contribute to research and prototype development in areas such as AI-assisted compiler development, kernel generation, RTL generation, hardware verification, and cross-layer system optimization. The role involves applying machine learning, generative AI, automated reasoning, and reinforcement learning to solve complex engineering challenges. Candidates will use technologies including Python, C, C++, Verilog, and SystemVerilog, while potentially engaging with LLVM, MLIR, Triton, CUDA, PyTorch, and JAX. The work focuses on improving end-to-end system performance and efficiency by connecting insights across the computing stack, specifically targeting the IBM Spyre accelerator software stack and IBM Z and Power systems.
Contribute to research and prototype development in areas like compiler development, kernel optimization, or RTL generation.
Apply machine learning, generative AI, or reinforcement learning to solve hardware and software engineering challenges.
Investigate post-training methods to extend the capabilities of open models for complex engineering tasks.
Design experiments and benchmarks to evaluate system performance, correctness, and engineering productivity.
Explore technical connections between specific project areas and adjacent layers of the computing stack.
Document methods and findings through technical presentations, reports, code, and potential research publications.
What we're looking for
Must have a Bachelor's Degree.
Must be currently enrolled in a Master's or PhD program in computer science, computer engineering, electrical engineering, or a related field (preferred).
Must possess strong programming or hardware-development skills in languages such as Python, C, C++, Verilog, or SystemVerilog.
Must have demonstrated experience in areas like compilers, kernel development, ML systems, reinforcement learning, computer architecture, or hardware-software co-design.
Must have experience conducting technical experiments, debugging complex systems, analyzing quantitative results, or developing research prototypes.
Must be able to communicate technical ideas and collaborate effectively in a multidisciplinary research environment.
Experience in reinforcement learning, generative AI, or LLMs for code and hardware design is beneficial (preferred).
Experience with frameworks like LLVM, MLIR, Triton, CUDA, PyTorch, or JAX is beneficial (preferred).
Data Science
Machine Learning
AI
Quantum Computing
High Performance Computing
Hybrid Cloud
Computer Science
Data Analysis
Semiconductor Research
Chip Architecture
Material Science
Physics
Chemistry
Mechanical Engineering
Python
C
C++
AI
Machine Learning
Compilers
LLM
CUDA
GPU Programming
LLVM
MLIR
Performance Engineering
Optimization
Data Structures
Algorithms
Reinforcement Learning
Predictive Modeling
Domain-Specific Languages