Harsha Vardhan Simhadri

Harsha Vardhan Simhadri is a Senior Principal Researcher at Microsoft, specializing in algorithmic research aimed at developing practical systems for large-scale applications. He has been with Microsoft since September 2016 and has played a pivotal role in several significant projects, particularly in the areas of approximate nearest neighbor search (ANNS) and machine learning for resource-constrained devices.

Professional Background

• Current Role: At Microsoft, Simhadri leads the DiskANN project, which focuses on creating efficient algorithms for vector search functionality. This project has been widely adopted across various Microsoft platforms, including web search, advertisements, and enterprise applications. DiskANN is notable for being the first external memory graph index for ANNS that supports real-time updates and streaming operations.12

• Previous Experience: Before joining Microsoft, he was a Postdoctoral Fellow at Lawrence Berkeley National Laboratory and completed his Ph.D. at Carnegie Mellon University, where he researched parallel algorithms and runtime systems. His educational background also includes a B.Tech in Computer Science from IIT Madras.134

Research Contributions

Simhadri's research is characterized by a focus on developing algorithms that bridge the gap between theoretical research and practical applications. Some of his notable contributions include:

• DiskANN: This project has revolutionized how vector search is implemented, allowing for the indexing of billions of points efficiently on commodity hardware. It has been recognized in various competitions, including the NeurIPS'21 challenge, where it demonstrated state-of-the-art performance.23

• EdgeML: He has also worked on machine learning operators and architectures designed for Internet of Things (IoT) devices, enhancing the capabilities of edge computing.14

• Mentorship: Beyond his research, Simhadri is active in mentoring research fellows and interns, fostering the next generation of researchers in the field.12

Personal Interests

Simhadri enjoys developing new algorithms motivated by real-world applications and is involved in organizing benchmarks for ANNS algorithms, contributing to the broader research community by creating datasets and hosting competitions.24