Prashant Vyas
- Role
- Engineering Leadership, Nvidia Gpu Cloud Ai Infra at NVIDIA
- Location
- Mountain View, CA, US
- LinkedIn followers
- 500 followers
Experience
Engineering Leadership, Nvidia Gpu Cloud Ai Infra
Jun 2017 — Present
Leading multiple distributed AI/ML workflows engineering within NGC AI Infrastructure. Leading the team of senior principal architects/engineers on multiple initiatives from the incubating, conceptualizing, architecting, executing these projects to build a production grade system and roadmap with the vision to solve these complex problems for NVIDIA- Incubated, conceptualized and executed the complex initiative on heterogeneous gpu/cpu resource management workflows to provide a fine-grained control over managing the pool of heterogeneous compute resources(bare-metal, vm) with policies to provide priority-based scheduling, hierarchical resource quota/fairness to enable elastic sharing of resources across organization- Incubated, conceptualized, executed multi-node Dask RAPIDS & Spark workflows from engineering skunkworks to a production grade system to provide a scalable distributed workflows offering for NVIDIA Base Command- Spearhead the strategic discussions and provided technical directions on MLOps initiatives representing NVIDIA to external partners. Identified and built core foundational components Secrets Manager for managing secrets to address strategic needs. Partnered with product management on leading Experiments Tracking, OAuth enabled APIs for Hyperparam Sweep- Incubated, architected and operated a multi-tier L4 load balancer using IPVS for the bare-metal deployment of multiple K8s Clusters and Object Storage providing seamless integration with K8s and standalone services on BMl nodes similar to AWS NLB K8s processing TBs/s- Hired top talent from top-tier companies and built squads of solid distributed systems leaders, architects/engineers- Architected an API Gateway, Service discovery and a federation for securely exposing services running inside the private clouds and federating AI batch workloads across heterogeneous GPU clusters- Launched NVIDIA AI Playground inference using NVIDIA Triton Inference on K8s.
Skills
- Rest
- Shell
- Web Services
- Shell Scripting
- Scalability
- Perl
- Php
- Mysql
- Apache
- Unix
- Linux
- Software Development
- C++
- C
- Distributed Systems
Find verified contacts for anyone on LinkedIn
Unifers gives sales teams verified emails and direct dials, enriched profiles, and outreach that lands in the inbox.
Free plan included · No credit card required
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.