Hi, I’m Ranjit.

I’m an infrastructure engineer with more than fifteen years across data centers, HPC, cloud and virtualization. My recent work is NVIDIA AI factory infrastructure, covering the whole stack: physical rack deployment and liquid cooling, GPU compute, high-performance networking, cluster provisioning, workload orchestration, storage and production operations.

That includes rack-scale GB200 and GB300 NVL72 systems and a 95-node NVIDIA DGX H200 SuperPOD. Before this I spent more than a decade designing and operating enterprise data centers in India, Germany and the United States. It’s why I treat an AI factory as one platform across compute, networking, storage, orchestration, power, cooling and operations, and not just a pile of GPUs.

• Deployed and brought up GB200 and GB300 NVL72 AI factory infrastructure, from rack installation through firmware validation and cluster acceptance testing to production readiness.
• Managed NVIDIA Base Command Manager, and configured and validated InfiniBand and Spectrum-X networking, including Cumulus Linux, BGP Unnumbered, VXLAN EVPN and NetQ.
• Supported direct-to-chip liquid cooling: CDU commissioning, manifold and quick-disconnect integration, leak testing and coordination with facilities.
• Ran GPU-enabled Kubernetes with Run:ai and Slurm HPC workloads, operated DDN parallel storage, and diagnosed InfiniBand and Ethernet problems affecting multi-node GPU jobs.
• Automated provisioning and validation with Ansible, Python and Bash, and built cluster observability with Prometheus and Grafana.

• 2025 to nowSr. Systems Engineer, AI Data Center and HPC Infrastructure
  The University of Texas System, Austin
• 2021 to 2024Senior HPC and Data Center Engineer
  Clario, Germany
• 2020 to 2021Deputy Manager, Cloud and Infrastructure Operations
  PHOENIX group, Germany
• 2019 to 2020Team Lead, Cloud and Infrastructure Operations
  adidas, Germany
• 2013 to 2019Senior Cloud Operations Engineer
  Vodafone, India
• 2010 to 2013Infrastructure Engineer
  ITC Infotech, India

AI Factory Fit is a personal portfolio project. It shows how I approach the first conversation with someone who says “I need 100 GPUs, a petabyte of storage and some HPC too.” Ask the right questions, size from sourced specifications, lay out the trade-offs, and be honest about what is an estimate.

Claude handles the conversation. The numbers come from a tested sizing engine and a data file where every input shows its source and whether it has been verified. Try the planner.

• Large-scale AI factory deployments
• GPU cloud infrastructure
• HPC platforms
• Next-generation data centers

B.Tech in Electrical and Electronics Engineering, Jawaharlal Nehru Technological University Hyderabad. AI & ML (Business Applications), The University of Texas at Austin.

NVIDIA-Certified Associate: AI Infrastructure and Operations. VMware Certified Professional, Data Center Virtualization (2023). VMware Cloud Foundation Administrator. VMware Certified Associate, Data Center Virtualization. EMC Data Domain Implementation Engineer.

I’m happy to talk about AI infrastructure, design questions or collaboration.

Show email addressLinkedIn

I reply from my personal address.