Sushma Vaddireddy

Software Engineer

San Francisco, California, United States5 yrs experience
Highly Stable

Key Highlights

  • Enhanced Performance Counters support by 300% on MI300.
  • Developed a profiling framework for HPC applications.
  • Achieved over 100x speed-up on GPUs for financial applications.
Stackforce AI infers this person is a Software Engineer specializing in HPC and Machine Learning applications.

Contact

Skills

Core Skills

Python (programming Language)C++DevopsCudaMpi

Other Skills

HIP ProgrammingrocProfDatabasesAutomationSQLJavaAngularOpenCVMachine LearningOpenMPRCCLNCCLJenkinsDocker ProductsGroovy

Experience

5 yrs
Total Experience
3 yrs
Average Tenure
2 yrs
Current Experience

Amd

2 roles

Senior Software Development Engineer

Jun 2024Present · 2 yrs · San Francisco Bay Area

HIP ProgrammingrocProfDatabasesAutomationSQLCUDA+14

Software Development Intern

May 2023May 2024 · 1 yr · San Francisco Bay Area

  • ▪ Extended a Python and HIP-based test suite to validate performance counters dumped by rocProf on MI300.
  • ▪ Developed/modified GPU kernels written in HIP to efficiently trigger performance counters by understanding GPU architecture and low-level assembly instructions.
  • ▪ Enhanced Performance Counters support from initial 48 to 197 (300%) on MI300.
C++Python (Programming Language)

Tata consultancy services

3 roles

System Engineer

Promoted

May 2020Jul 2022 · 2 yrs 2 mos · Hyderabad, Telangana, India

  • My responsibilities while working with AMD ROCm project:
  • Developed a Python framework on top of the rocProf profiler to automatically collect and evaluate profiling data for HPC applications across various domains such as Molecular Dynamics, Quantum Mechanics, Hydrodynamics.
  • Extended framework capabilities to include profiling of ML Frameworks like Tensorflow, Pytorch.
  • Added functionality to store the profiling data in a Django database and automatically send the data via email using the MIME module.
  • Developed a groovy script for Jenkins server to provide continuous integration/development, which automates the entire process of loading the latest committed code and evaluating the profiling data of HPC applications/ML frameworks
DevOpsGroovy

Assistant System Engineer

Jun 2019Apr 2020 · 10 mos · Hyderabad, Telangana, India

  • My responsibilities while working with TCS R&D Team:
  • Worked on parallelizing and scaling of ”Stock Option Pricing” application to predict the ”put” and ”call” options for a stock on multi-node CPUs and GPUs using Monte Carlo Algorithm.
  • Parallelized the application on multi-core CPUs using OpenMP and accelerated it on multi-node GPUs using CUDA, MPI, and NCCL, which resulted in a speed-up of over 100 times on GPUs and 60 times on CPUs, compared to the open-source implementations.
MPIOpenMPCUDA

Project Intern

Jan 2019May 2019 · 4 mos · Hyderabad, Telangana, India

  • Worked on Suspicious Activity Detection project mentored by TCS Nextgen R&D Team:
  • Developed a Computer vision-based Deep Learning solution for reporting suspicious activity, including weapons, fire, fighting scenes in public places, to the concerned authorities by analyzing CCTV footage.
Python (Programming Language)OpenCV

Education

University of Southern California

Master of Science - MS — Computer Science

Aug 2022May 2024

Gokaraju Rangaraju Institute of Engineering and Technology

Bachelor of Technology - BTech — Computer Science

Aug 2015May 2019

Sri Gayathri Junior College

Intermediate — MPC

Jun 2013Apr 2015

Krishnaveni Talent School - India

SSC (10th Standard)

Stackforce found 100+ more professionals with Python (programming Language) & C++

Explore similar profiles based on matching skills and experience