AI Systems Research Intern — Agentic AI for Z

IBM; Cambridge (MA); On-site; USD 89,000 - 164,000; Full time

Job summary

IBM Research seeks a PhD student for a 2027 summer internship to advance agentic AI systems for IBM Z and Spyre-enabled environments. You will explore techniques for efficient information control, memory management, and evidence-based reasoning within the Finn/Paver framework.

The role focuses on Python-based prototyping, experimental design, and collaboration with IBM researchers and academia to publish results or contribute IP disclosures. Strong ML and NLP background is essential.

Qualifications

• Pursuing a PhD in CS/CE/EE/AI or related field.
• Experience in ML, deep learning, LLMs, AI agents, NLP.
• Strong Python programming with ML frameworks (PyTorch, TensorFlow).
• Experience designing experiments and prototyping research.
• Excellent written and verbal communication skills.

Responsibilities

• Research, design, and prototype adaptive information-control mechanisms for agentic AI systems in IBM Z environments.
• Model minimal sufficient agent state including observations, evidence, memory, and task progress.
• Develop techniques for information acquisition, retrieval, summarization, and memory management.
• Implement experimental solutions in Python and integrate into Finn/Paver or similar agents.
• Design experiments on z/OS management tasks measuring reliability, memory use, and latency.
• Analyze tradeoffs in performance, information efficiency, and resource requirements for on-platform execution.
• Collaborate with researchers to review results and translate findings into architectures.
• Document outcomes through reports, presentations, and potential publications.

Skills

Python

ML research

LLMs

Distributed systems

Communication

Education

Master's Degree

Doctorate Degree

Tools

PyTorch

TensorFlow

Hugging Face

Job description

IBM Research seeks a PhD student for a 2027 summer internship to advance agentic AI systems for IBM Z and Spyre-enabled environments. You will explore techniques for efficient information control, memory management, and evidence-based reasoning within the Finn/Paver framework.

The role focuses on Python-based prototyping, experimental design, and collaboration with IBM researchers and academia to publish results or contribute IP disclosures. Strong ML and NLP background is essential.