Deep Shekhar

Software Engineer

Bengaluru, Karnataka, India7 yrs 6 mos experience

Most Likely To SwitchAI Enabled

Key Highlights

Achieved significant performance improvements in DLSS 3.0.
Core engineer for innovative AI gaming project G-Assist.
Expertise in GPU performance optimization for AAA games.

Stackforce AI infers this person is a GPU Performance Architect specializing in gaming and AI applications.

Contact

shekhard@nvidia.com +917427980024 LinkedIn

Skills

Core Skills

GpuPerformance OptimizationGenerative AiCudaComputer Vision

Other Skills

AAA GamesAI for GamingAlgorithmsAnomaly DetectionBashC++CommunicationComputer ArchitectureComputer GraphicsDLSS 3.0Data AnalysisDeep LearningDigital Image ProcessingDirectXGPU Simulator

About

Experienced GPU architect and performance engineer at Nvidia, specializing in hardware/software optimizations for AAA games and AI workloads. Working to design, implement, and optimize cutting-edge AI for graphics applications, driving innovation and performance improvements across the PC graphics ecosystem.

Experience

7 yrs 6 mos

Total Experience

3 yrs 9 mos

Average Tenure

6 yrs 11 mos

Current Experience

Nvidia

2 roles

Senior Architect

Promoted

Jun 2022 – Present · 4 yrs

Performance optimization effort for DLSS 3.0:
Characterized the performance bottlenecks and quantized the same using the .
Demonstrated ~10% perf-upside on the DLSS3.0 through source code optimizations, some of which were taken into production.
Prototyped multi-viewport rendering with ultra settings with minimal FPS impact. Achieved via conjunction of viewport specific dynamic refresh rate and DLFG.
AI for Gaming Projects:
Auto generate context aware and engaging(e.g. color commentary) descriptions of gameplay
One of the core engineers of Nvidia's Project G-Assist, Link: https://www.nvidia.com/en-in/geforce/news/g-assist-ai-assistant/

C++GPUAI for GamingPerformance OptimizationDLSS 3.0

GPU Performance Architect

Jul 2019 – Jun 2022 · 2 yrs 11 mos

Worked on mitigating performance bottlenecks for AAA pc games on latest Nvidia GPUs.
Devised SW/HW strategies, as well as applied previously created optimizations to speed up AAA PC Games by 3-10% on latest Nvidia GPUs.
Ported traditionally serial workloads to Nvidia GPUs using CUDA. Exploited potential parallelism in those workloads to achieve >10x speedup vs the modern CPUs.

CUDAPerformance OptimizationAAA GamesSW/HW Strategies

Indian institute of technology, kharagpur

Graduation Project

Sep 2018 – Apr 2019 · 7 mos · India

Worked on enabling a Multitasking surveillance Bot under Dr. Debashis Sen, Assistant Professor at department of Electronics and Electrical Communication, IIT Kharagpur.
Implemented on a TurtleBot. Object Detection and Inverse Perspective Algorithm which captured the objects of interest and their relative position. The system also consists of extracting color and shape feature from objects of interest and finally an Anomaly Detection Model is trained for our environment which allows Bot to report Unknown objects.

Object DetectionAnomaly DetectionInverse Perspective AlgorithmComputer Vision

Nvidia

Summer Intern

May 2018 – Jul 2018 · 2 mos · Bengaluru, Karnataka, India

Investigated better cache utilization opportunities on GeForce GPUs for various Graphics and Compute workloads.
Worked on C++ based GPU Simulator and Python Scripts during the internship.
Key contributions during the internship :-
[1] Discovering and Analyzing the presence of the redundant data in Texture(L1 level) Caches and it's impact on the GPU performance.
[2] Worked on ideas to curb the effects of the redundant data which would eventually improve performance.

C++PythonGPU SimulatorGPU