Summary
Posted: Oct 09, Weekly Hours: Role Number: Application Deadline: Sunday 16th November We are looking for a machine learning research intern with passion to spearhead new solutions to challenging problems in foundation model development with an emphasis on agentic capabilities and Reinforcement Learning. You will work closely with an applied ML team of around 30 engineers in Zurich, and get the chance to connect with other experts in the US
Description
During your internship, you will do research and development in the context of Apple foundation models that power next-generation AI experiences across Apple's platforms. You will learn what is relevant to transfer results from machine learning research to meaningful real-world performance at scale. Your work can contribute to Apple Intelligence and be shipped to billions of devices, worldwide
Responsibilities
* Research and develop novel approaches for improving foundation model capabilities
* Leverage brand new, large-scale RL methods to improve reasoning and planning, multi-turn conversations, and general agentic capabilities.
* Design and train agents with tool calling, planning, and API integration to successfully complete tasks.
* Investigate multimodal foundation models spanning text, vision, audio, and other modalities
* Collaborate on model alignment, safety, and responsible AI practices
* Contribute to the development of tools and frameworks for foundation model experimentation and evaluation
Minimum Qualifications
* PhD student in Machine Learning (CS, ECE, Statistics, Math, natural sciences, or other related fields)
* Strong background in Reinforcement Learning and Deep Learning, with hands-on experience training large models, in particular LLMs
* Experience with training methodologies including pre-training, fine-tuning, and alignment approaches
* Strong programming skills. Most relevant are Python, PyTorch, and JAX
Preferred Qualifications
* Excellent problem solving, critical thinking, and interpersonal skills.
* Ability to define and drive ambitious goal-oriented research and development independently