GPU Performance Modeling Engineer

Minimum Qualifications:

• Bachelor’s degree or equivalent practical experience.
• 2 years of experience with software development in one or more programming languages, or 1 year of experience with an advanced degree.

Preferred Qualifications:

• Master's degree or PhD in Computer Science or related technical fields.
• 3 years of experience with LLMs/ML, algorithms and tools (e.g. TensorFlow/Jax), Artificial Intelligence, deep learning, or natural language processing.
• 2 years of experience with data structures or algorithms in either an academic or industry setting.
• 2 years of experience building and developing large-scale infrastructure, distributed systems or networks, or with compute technologies, storage, or hardware architecture.
• Experience in developing and deploying AI/ML models and algorithms and with Python and any other languages (e.g., C++, Kotlin, Java.).
• Understanding of ML, data analysis and developer tools.

About The Job

Google's software engineers develop the next-generation technologies that change how billions of users connect, explore, and interact with information and one another. Our products need to handle information at massive scale, and extend well beyond web search. We're looking for engineers who bring fresh ideas from all areas, including information retrieval, distributed computing, large-scale system design, networking and data storage, security, artificial intelligence, natural language processing, UI design and mobile; the list goes on and is growing every day. As a software engineer, you will work on a specific project critical to Google’s needs with opportunities to switch teams and projects as you and our fast-paced business grow and evolve. We need our engineers to be versatile, display leadership qualities and be enthusiastic to take on new problems across the full-stack as we continue to push technology forward.

Graphics Processing Unit (GPU) Performance team is responsible for optimizing, modeling and evaluating GPU systems for comparative analysis and benchmarking for Google’s internal machine learning (ML) workloads. We strive for extracting maximum efficiency in Google’s GPU fleet. The team’s focus on performance analysis and optimization identifies opportunities in Google production and research ML workloads and land optimization to the entire fleet. The team evaluate current and future ML workloads and run Performance over Total Cost of Ownership (TCO) simulations to collect roofline estimates and guide decision making for the Cloud hardware teams.

The ML, Systems, & Cloud AI (MSCA) organization at Google designs, implements, and manages the hardware, software, machine learning, and systems infrastructure for all Google services (Search, YouTube, etc.) and Google Cloud. Our end users are Googlers, Cloud customers and the billions of people who use Google services around the world.

We prioritize security, efficiency, and reliability across everything we do - from developing our latest TPUs to running a global network, while driving towards shaping the future of hyperscale computing. Our global impact spans software and hardware, including Google Cloud’s Vertex AI, the leading AI platform for bringing Gemini models to enterprise customers.

The US base salary range for this full-time position is $141,000-$202,000 + bonus + equity + benefits. Our salary ranges are determined by role, level, and location. Within the range, individual pay is determined by work location and additional factors, including job-related skills, experience, and relevant education or training. Your recruiter can share more about the specific salary range for your preferred location during the hiring process. Please note that the compensation details listed in US role postings reflect the base salary only, and do not include bonus, equity, or benefits. Learn more about benefits at Google.

Responsibilities

• Identify and maintain Large Language Model (LLM) training and serving benchmarks that are representative to Google production, industry and ML community, use them to identify performance opportunities and drive XLA:GPU/Triton  performance toward state-of-the-art, and to guide XLA releases. 
• Engage with Google product teams like Deepmind to solve their ML model performance problems, for example, onboarding new LLM models and products on GPU hardware, enabling LLMs to train and serve efficiently on a very large scale (i.e., thousands of GPUs).
• Run architecture level simulations on GPU designs and perform roofline analysis to guide internal teams.
• Analyze performance and efficiency metrics to identify bottlenecks, design and implement solutions at Google fleetwide scale. Run performance benchmarks on GPU hardware using internal and external tools.