Research scholar program

Our team conducts research in graph mining, optimization, operations research, and market algorithms to improve Google's infrastructure, machine learning, and marketplaces. We collaborate with teams across Google and perform research in related areas, such as algorithmic foundations of machine learning, distributed optimization, economics, and data mining.

Large language, visual, and multimodal models have made significant advances in recent years, opening up new possibilities for scientific research. We invite proposals in these four areas:

• Applications: Proposals that demonstrate how large language models can be used to advance scientific discovery in a specific field.
• Foundations: Proposals that explore broad advances in building, tuning, or deploying large models for scientific research, such as integrating language models with specialized scientific tools, developing multimodal models for understanding scientific data, and accelerating scientific analysis, experimentation, and summarization.
• Evaluation: Proposals that develop datasets or methods for benchmarking and evaluating large models for science, including evaluating domain-specific knowledge, assessing factuality and grounding, evaluating multimodal capabilities, and developing tasks that require multi-step scientific reasoning.
• HCI: Proposals that enhance scientific workflows, such as automating complex simulation pipelines, with large language models and human-in-the-loop interaction.

Google Health research aims to advance AI and technology to help people live healthier lives through collaborative research with public officials, clinicians, and consumers. We are developing tools to understand population health, novel algorithms to better understand and use complex medical data, and technology to help people find high-quality health information and understand their health status.

We invite proposals that will generate and understand large datasets to improve population health, develop novel algorithms for better understanding of complex medical data, and develop novel methods to extract health insights cheaper, faster, or better.

HCI researchers at Google design and build large-scale interactive systems that aim to be humane, simple-to-understand, and delightful to use. We work across a variety of HCI disciplines, including predictive and intelligent user interfaces, mobile and ubiquitous computing, social and collaborative computing, and interactive visualization.

Machine learning is the foundation of Google's research, with a broad scope that includes foundational and algorithmic work, critical real-world applications, and topics, such as federated learning, information retrieval, learning theory, optimization, reinforcement learning, robotics, and recommender systems.

Machine perception researchers at Google develop algorithms and systems to tackle a wide range of tasks, including action recognition, object recognition and detection, hand-writing recognition, audio understanding, perceptual similarity measures, and image and video compression.

Our team comprises multiple research groups working on a wide range of natural language understanding and generation projects. Our researchers are focused on advancing the state of the art in natural language technologies and accelerating adoption everywhere for the benefit of the user. Natural language processing and understanding plays a major role in driving Google’s company-wide OKRs as language understanding is the key to unlocking Google’s approach: “Build a more helpful Google for everyone that increases the world’s knowledge, success, health, and happiness.”

Google's privacy research reaches across multiple teams, focusing on different aspects of privacy to advance the state of the art and develop tools to protect users and give them control over their data. This includes work on privacy-preserving technologies using cryptography and differential privacy, machine learning for privacy, user interface design and human-computer interactions to make communication clear and empower users, privacy policy to define Google's guiding principles for user protection, and system analysis and measurement to develop techniques to evaluate the privacy health of Google's systems.

The Quantum AI team is developing an error-corrected quantum computer and discovering valuable applications by offering a quantum computing service. We collaborate with academic partners to advance both goals, so if you have a quantum algorithm you would like to run on our service, please submit a proposal.

Google's security and anti-abuse research team brings together experts from multiple disciplines to defend users from a wide range of threats. This includes work on access control, information security, networking, operating systems, language design, cryptography, fraud detection, machine learning for abuse detection, denial of service, emerging threats, user interfaces, and other human-centered aspects of security.

As Moore's Law slows down, we enter a new era of "specialization," where hardware and software engineers must collaborate more effectively across their domains. This necessitates shared artifacts, mutual understanding of cost models and collaborative tooling/methodology. Google seeks to support academic research that utilizes automation, software engineering practices and machine learning to accelerate this co-design process.

Our program specifically calls for proposals in the following areas:

• Optimization of mixed cell heights in advanced semiconductor nodes

• Exploration of methods to apply machine learning for delay modeling in high-level synthesis tooling

• Development and contribution to a corpus of retargetable hardware libraries, such as RISC-V

• Investigation of concolic testing applications for dynamic coverage collection in high-level synthesis tooling

Google's systems and networking systems research is focused on building and deploying novel systems at unprecedented scale. Our work spans the entire spectrum of computing, from large-scale distributed systems to individual machines to accelerator technologies.

We address fundamental questions around data center architecture, cloud virtual networking, wide-area network interconnects, software-defined networking, machine learning for networking, large-scale management infrastructure, congestion control, bandwidth management, capacity planning, and designing networks to meet traffic demands.