Technical leader with a decade of experience engineering Google's network infrastructure, spanning Data center, WAN and AI fabrics. My focus is on incubating and driving ground-up innovations that are critical to the success of our key customers and internal services and differentiate Google's AI infrastructure. I've driven optimizations to resolve critical performance bottlenecks in Google's AI networking stack, enabling strategic partners like Anthropic, LG AIR, Magic and many others to maximize workload performance on GCP.

I build and scale systems across the full networking stack - from hardware pipelines (NIC and Switch architectures), host & switch networking stacks (kernel & userspace) and transports (congestion control, load balancing, RDMA), to large-scale distributed systems underpinning SDN control planes for telemetry, traffic engineering, scheduling, bandwidth allocation & capacity planning.

I'm the co-inventor and overall lead for Congestion Signaling (CSIG), a technology that redefines in-band network telemetry and traffic management for AI fabrics in the industry. Over five years, I took CSIG from a concept to at-scale production deployment at Google, leading a cross-functional effort of 30+ talented engineers & researchers. I also drove its adoption as an industry standard through engagements with Ultra-Ethernet Consortium (UEC), OCP, IETF and IEEE, leading to its adoption by 7+ switch/NIC vendors and native HW support for the CSIG protocol in over a dozen market-leading ASICs.

During my final year, I helped teach a couple of courses:

CS2310 - Digital Logic and Design Lab: Led a team of seven graduate teaching assistants, structured the course and designed assignments in addition to helping with and evaluating students’ progress.

CS3210 - Computer Networks Lab: Led a team of ten teaching assistants, taught classes and designed assignments. Introduced the class to the GENI Networking test-bed (As part of the GENIForEducation initiative). The lab was the first undergraduate course in India to feature assignments on GENI .

Worked on enhancements to Information-Centric Networking, a next-generation Internet architecture, to improve content access latency and to demonstrate the benefits of pervasive router-level caching.

• Built a multiplayer network architecture to support collaborative actions in virtual reality over

gigabit-speed networks - Integrated the architecture into a virtual reality simulator for

orthopaedic surgery training.

• The application was recognized as one among the top 30 next-generation applications in the United States and was demoed at US-Ignite in Sunnyvale, CA and also at GEC20, the 20th GENI Engineering Conference in 2014.