2027 Spring Intern - Data Science & Machine Learning

Roche•South San Francisco, CA

6d•Onsite

About The Position

Genentech is seeking a Spring intern to join its Data Science & Machine Learning team and help research, prototype, and evaluate Agentic AI systems. The work spans Generative AI, software prototyping, and system evaluation, including specialized question-answering agents and multi-agent evaluation frameworks. A central focus is ensuring systems are safe, reliable, compliant, and aligned with healthcare standards. The intern will build and refine AI-agent workflow prototypes for interactive question-answering, search, and workflow automation; develop an evaluation framework for LLM outputs and agent performance; and implement safeguards for patient safety, privacy, and regulatory compliance. The role also involves writing maintainable Python code to connect LLMs with internal APIs, search tools, and structured data, collaborating with cross-functional partners, and presenting project findings and metrics at the end of the internship. This is an intensive, paid, 12-week, full-time internship of 40 hours per week, with the possibility of extension to six months or one year. The program begins in January. The position is based on-site in South San Francisco, California. A location-based stipend is provided. The stated pay is $52 per hour, and the position qualifies for paid holiday time off. Relocation benefits are not available.

Requirements

Must be an enrolled student pursuing a master's degree or PhD.
Must be pursuing a degree in Computer Science, Data Science, AI/ML, Biomedical Informatics, or a related quantitative field.
Foundational proficiency in Python and standard software development practices, including version control/Git and modular code.
Understanding of Large Language Models and core Generative AI patterns, such as prompt engineering, agentic workflows, or Retrieval-Augmented Generation.
Awareness of AI safety principles, output reliability, and the importance of compliance in high-stakes fields such as biotechnology and healthcare.

Nice To Haves

Excellent communication, collaboration, and interpersonal skills.
Alignment with the company's values of Integrity, Courage, and Passion.
Academic projects or coursework involving LLM API integration or open-source AI frameworks such as LangChain or LlamaIndex, or experience building with commercial or open-source APIs.
Experience or strong interest in system evaluation, benchmarking, or testing, such as creating test cases, evaluating model accuracy or consistency, or unit testing agent steps.
Strong analytical problem-solving skills, curiosity, and eagerness to experiment in a fast-paced environment.

Responsibilities

Build, test, and refine functional AI-agent workflow prototypes for interactive question-answering, search, and workflow automation.
Design and implement a general-purpose framework to benchmark LLM outputs, agent reliability, tool calling, and reasoning accuracy.
Explore and implement safeguards to make agent outputs safe, resistant to hallucinations, compliant with regulatory standards, and aligned with patient safety and privacy guidelines.
Develop clean, maintainable Python code integrating LLMs with internal APIs, search tools, and structured data sources.
Collaborate with engineers, data scientists, and domain stakeholders to gather requirements, incorporate feedback, and demonstrate prototype capabilities.
Present project findings, system architectures, and evaluation metrics to the broader team at the end of the internship.