Krishnakant Jaju

Director of Engineering

Pune, Maharashtra, India13 yrs 10 mos experience

Highly Stable

Key Highlights

Expert in virtualization and device drivers.
Led development of NVIDIA's first SR IOV capable chip.
Significant contributions to vGPU live migration architecture.

Stackforce AI infers this person is a skilled Software Engineer specializing in virtualization and system software for infrastructure solutions.

Contact

Skills

Core Skills

VirtualizationDevice Drivers

Other Skills

Data CentersCloud ComputingProgrammingLinuxCData StructuresShell ScriptingC++PythonAlgorithmsPerforceKernel ProgrammingOperating SystemsXenKernel Debugging

About

Software Engineer with a demonstrated history of working in the system software domain. Skilled in Virtualization, Linux, Device Drivers, Data Structures and Algorithms

Experience

13 yrs 10 mos

Total Experience

13 yrs 10 mos

Average Tenure

13 yrs 10 mos

Current Experience

Nvidia

3 roles

Software Engineering Manager

Promoted

Sep 2022 – Present · 3 yrs 8 mos

Senior System Software Engineer

Oct 2016 – Aug 2024 · 7 yrs 10 mos

NVIDIA vGPU (Virtual GPU) – SRIOV vGPU
● Worked on enabling the first SR IOV capable NVIDIA chip based on Ampere architecture.
● Implemented dirty page tracking solution for guest system memory used by SR IOV capable vGPUs.
● Worked on adding MIG (Multi-Instance GPU) support on NVIDIA A100 board running vGPU software and integrated MIG management with Hypervisor and vGPU Manager.
NVIDIA vGPU (Virtual GPU) – Live Migration
● Involved in design and implementation of vGPU Live Migration architecture.
● Supporting vGPU live migration required adding new communication and notification interfaces between VMM and vGPU. Designed these interfaces in collaboration with Hypervisor vendors.
● Worked on save and restore of the vGPU device state (SW and HW) and compatibility algorithms between source and destination.
● Implemented selective GPU Frame Buffer copy algorithm to reduce Frame buffer memory transfer during Live Migration.
NVIDIA vGPU (Virtual GPU) – vGPU Management
● Implemented backend of NVIDIA GRID vGPU Management.
● Developed a system service which reads predefined vGPU types, with fixed configurations of framebuffer, display heads and other capabilities, from a xml file and writes them to NVIDIA kernel driver.
● Extended NVIDIA management tools to support vGPU and develop a C-based SDK for monitoring and managing various states of NVIDIA GRID vGPUs.

Data CentersCloud ComputingVirtualizationDevice Drivers

System Software Engineer

Jun 2012 – Sep 2016 · 4 yrs 3 mos

Worked on NVIDIA's Windows and Linux Device Drivers, to improve the effective performance of the Virtual Machines and to increase the maximum number of VMs that can run vGPU with acceptable performance on a single GPU.
Designed and Implemented communication channel between guest OS and host, using shared memory mechanism and remote procedure calls (RPCs.)
Major contribution in increasing the host stability by reducing dynamic memory footprint for VMs and judiciously using GPU BAR regions.