Machine Learning Ph.D Student - Agentic AI

Описание

As a member of Apple's ML group, you will design, prototype, and ship agentic GenAI systems used by chip design engineers to accelerate their day-to-day workflows. This role involves taking LLMs from "promising demo" to "depended-on tool" by designing agent loops, retrieval pipelines, and evaluation frameworks.

Responsibilities

• Design and build agentic systems featuring multi-agent coordination, long-horizon reasoning, tool use, and planning frameworks.
• Develop end-to-end solutions, including domain-specific agents and high-performance retrieval pipelines spanning diverse data sources.
• Partner with subject-matter experts on the customer teams to translate their workflows and rules of thumb into concrete agent assignments, prompts, tools, and guidelines.
• Build the evaluation backbone for agents: define success measures, curate offline golden sets, and instrument online telemetry to power feedback loops that drive fast, safe iteration with real cost and quality control.
• Drive prompt engineering, model selection, and lightweight model adaptation as the workloads demand.
• Collaborate closely with software engineers on the team to productionize your work, and with researchers and customers to shape what is built next.
• Stay current with the rapidly evolving agentic and LLM landscape and bring promising techniques into products with good engineering judgment.

Requirements

• MSc, or PhD in Computer Science, Computer Engineering, Electrical Engineering, or related fields.
• Strong programming skills in Python and experience with modern ML and deep learning frameworks (e.g., PyTorch).
• Hands-on experience with agentic frameworks, tool use, and data pipelines.
• Strong evaluation mindset: comfortable designing golden sets, metrics, and telemetry to drive iteration.
• Excellent communication and collaboration skills, with the ability to align with subject-matter experts and translate their domain knowledge into agent behavior.

Nice to have

• Hands-on experience with stateful, multi-turn agentic frameworks and complex execution flows.
• Experience with prompt engineering and optimization techniques.
• Experience with model adaptation (e.g., fine-tuning, distillation, preference optimization).
• Publications at top-tier venues (e.g., NeurIPS, ICML, ICLR, ACL, EMNLP, or similar).
• Familiarity with chip design or EDA environments.

We offer

• Apple believes accessibility is a fundamental human right, reflected in its culture, benefits and digital tools.