AI Native Hardware & Software Codesign internship 2027

IBM; San Jose (CA); On-site; USD 36,000 - 54,000; Full time

Job summary

IBM Research AI Native Systems is seeking an AI-Native Hardware-Software Co-design Research Intern in San Jose. You will join a multidisciplinary team to advance AI-native engineering methods, contribute to compiler generation, RTL/verification, and cross-layer optimization, and help extend open models for hardware tasks.

Ideal candidates are pursuing a Master’s or PhD in CS/CE/EE with strong programming or hardware skills and a passion for research.

Qualifications

• Enrolled in a Master’s or PhD program in CS/CE/EE or closely related field.
• Strong programming or hardware-development skills (Python, C, C++, Verilog, SystemVerilog).
• Experience in at least one area: compilers, kernel development, ML systems, RL, architecture, or hardware verification.

Responsibilities

• Contribute to research and prototype development in one or more project areas.
• Apply ML, generative AI, automated reasoning, RL, or systems techniques to hardware/software challenges.
• Design experiments and benchmarks to evaluate correctness, performance, or efficiency.
• Collaborate across AI, compilers, architecture, and systems software teams.

Skills

Python

C/C++

Verilog/SystemVerilog

Machine learning

Research

Education

Master's degree program in CS/CE/EE

Tools

LLVM

MLIR

Triton

CUDA

PyTorch/JAX

Job description

Introduction

AI is changing both the systems we build and the way we conduct research and engineering. Within IBM Research AI Native Systems, our mission is to position IBM at the forefront of enterprise AI deployment while transforming how future systems are designed and built.

Introduction

AI is changing both the systems we build and the way we conduct research and engineering. Within IBM Research AI Native Systems, our mission is to position IBM at the forefront of enterprise AI deployment while transforming how future systems are designed and built. The AI-Native Hardware-Software Co-design Methodologies organization brings hardware and software together through a continuous co-design loop. Our research portfolio spans AI-assisted compiler development, kernel generation and optimization, RTL generation, hardware implementation, verification, and cross-layer system optimization. We also investigate reinforcement learning and other post-training techniques that can extend open models to solve increasingly difficult hardware and software engineering problems. Interns join focused projects within this broader research portfolio based on their experience and interests. By connecting insights across teams and technical layers, we aim to shorten development cycles, improve end-to-end system performance and efficiency, and uncover optimization opportunities that may be missed when hardware and software are developed independently. This internship offers the opportunity to work with a multidisciplinary research team, contribute to emerging AI-native engineering methods, and explore ideas that can shape future IBM systems and the broader computing ecosystem.

Your Role And Responsibilities

As an AI-Native Hardware-Software Co-design Research Intern, you will join a project aligned with your technical background and interests. Project areas may include AI-assisted compiler development, kernel generation and optimization, RTL generation, hardware verification, cross-layer hardware-software optimization, or reinforcement learning methods that extend open models to address challenging engineering problems. You are not expected to have experience across all these areas.

Working with a research mentor and an interdisciplinary team, you will:

• Contribute to research and prototype development in one or more project areas aligned with your expertise.


• Apply machine learning, generative AI, automated reasoning, reinforcement learning, or systems techniques to a well-defined hardware or software engineering challenge.


• Depending on the project, investigate reinforcement learning and related post-training methods for extending the capabilities of open models on complex tasks such as RTL generation, optimization, and verification.


• Design experiments and benchmarks to evaluate correctness, performance, efficiency, engineering productivity, or development-cycle improvements.


• Explore connections between your assigned area and adjacent layers of the computing stack.


• Collaborate with researchers and engineers across AI, compilers, computer architecture, digital design, verification, and systems software.


• Document your methods and findings and communicate results through technical presentations, reports, code, and potential research publications.

Projects may involve the IBM Spyre accelerator software stack and IBM Systems, including IBM Z and IBM Power, while also exploring methods applicable to future processors, accelerators, and computing platforms.

Preferred Education

Master's Degree

Required Technical And Professional Expertise

• Currently enrolled in a Master's or PhD program in computer science, computer engineering, electrical engineering, or a closely related field.


• Strong programming or hardware-development skills using languages relevant to the candidate's area of expertise, such as Python, C, C++, Verilog, or SystemVerilog.


• Demonstrated experience in at least one of the following areas: compilers, kernel development, machine learning systems, reinforcement learning, computer architecture, digital design, RTL development, hardware verification, or hardware-software co-design.


• Experience conducting technical experiments, debugging complex systems, analyzing quantitative results, or developing research prototypes.


• Ability to communicate technical ideas and collaborate effectively in a multidisciplinary research environment.A44

Preferred Technical And Professional Experience

Experience in one or more of the following areas is beneficial. Candidates are not expected to meet every preferred qualification.

• Reinforcement learning, reward modeling, model evaluation, or related post-training techniques for extending open models to address complex reasoning, code-generation, or engineering tasks.


• Applying generative AI, large language models, program synthesis, or automated reasoning to code, systems, hardware design, or verification.


• Compiler or kernel technologies such as LLVM, MLIR, Triton, CUDA, PyTorch, JAX, or related frameworks.


• Verilog or SystemVerilog development, hardware simulation, synthesis, formal verification, or electronic design automation tools.


• AI accelerators, heterogeneous computing, performance modeling, workload characterization, or cross-layer hardware-software optimization.


• Research demonstrated through publications, technical projects, open-source contributions, previous internships, or advanced coursework.