About
Portrait of Bhuvan Prakash Karuturi

Bhuvan Prakash Karuturi

I build the compiler, runtime, and virtualization stack behind Google Cloud TPUs.

About Me

Hi, I'm Bhuvan Prakash Karuturi, a Senior Software Engineer at Google with 5+ years of professional experience in ML Infrastructure and Accelerated Computing. I research, design, and develop high-performance software for computing platforms, with expertise in compilers, runtimes, virtualization stacks, and system-level performance optimization. My work focuses on scaling machine learning workloads through hardware-software co-design and efficient distributed systems.

Skills

ML Infrastructure & Platforms

  • Accelerated Computing
  • ML Compilers & Runtimes
  • HW-SW Co-design
  • Systems Performance & Profiling
  • Virtualization & Guest Environments
  • System Benchmarking

Systems & Backend

  • Distributed Systems
  • Operating Systems
  • Linux Kernel & Internals
  • REST & GraphQL
  • Node.js & Express.js
  • BigQuery & SQL

Languages

  • C++
  • Go
  • Python
  • Java
  • JavaScript / TypeScript

Tools & DevOps

  • Kubernetes
  • Docker
  • CI/CD
  • Git & GitHub
  • Angular & React
  • Redux

Work Experience

Google LLC

Senior Software Engineer

Greater Seattle Area, Washington

Jan 2022 – Present

  • Tech lead for the Cloud TPU7 and TPU v5p new-product introductions, and contributor to several other TPU launches.
  • Designed and built a novel simulation system that accelerated TPU new-product introductions.
  • Develop the Cloud TPU compiler and runtime stack, virtualization stack, guest environment, and the TPU compiler/runtime SDK.
  • Own end-to-end system benchmarking and profiling for the Cloud TPU product.

Amazon.com Services LLC

Software Development Engineer Intern

Seattle, WA

May 2021 – Aug 2021

  • Developed an ML-based system to classify fraudulent transactions based on features derived from 4 different systems. Gained experience of working with terabyte-scale data in various data stores such as RedShift, DynamoDB.
  • Trained an ML model which demonstrated the ability to reduce the labor costs for manual review by 73%.

Ohio-At-Home Healthcare Agency

Full Stack Web Developer Intern

Columbus, OH

Aug 2020 – May 2021

  • Developed the mobile-first provider portal frontend using Angular 10 and Material Design. This is the primary client interface and has been in use since 2021.
  • Developed 6 microservices to process business logic. Used asynchronous communication between services using Azure Event Bus message queues.

Capgemini India

Senior Software Engineer

Mumbai, India

May 2017 – May 2019

  • Customized an ERP system by building robust modules of Java, SQL, PL/SQL that received faster user acceptance.
  • Received the Project STAR Award in 2018 for prompt and efficient solutions delivered as part of the engagement.

Education

The University of Texas at Dallas

Master of Science in Computer Science

Dallas, Texas

Aug 2019 – May 2021

Acharya Nagarjuna University

Bachelor of Technology in Electrical Engineering

Guntur, India

Sep 2013 – Apr 2017

Earlier Projects

JavaDB — MySQL-like Database in Java

  • Built a MySQL-like database management system from scratch in Java, implementing full CRUD functionality.
  • Implemented search optimization by creating indexes using B and B+ trees.

Filesystem Consistency Checker for MIT xv6

  • Built a filesystem consistency checker in C for the filesystem used by the xv6 operating system.
  • Implemented 13 different consistency checks involving inodes, bitmaps, data blocks, and directories.

Classroom Management with POSIX Threads in C

  • Built a mock classroom management system with student, tutor, and coordinator threads to achieve parallelism.
  • Managed resource allocation with semaphores and mutex locks to avoid deadlocks.

More web projects

SocialApp — Social Media Web Application

  • Developed a social media application using Firebase cloud services, Node.js, React, Redux, and Material-UI.
  • Used SOA and microservices to expand compatibility and to halve the frontend package size and billing costs.

ContactList — Full Stack Web Application

  • Developed a RESTful web app with a Java Spring backend and a jQuery frontend.
  • Used JPA through Hibernate to communicate with MySQL and Spring Web as the REST controller.