Pavani Majety's Resume

June 2022 - Present

Contributed to Tensorflow-TensorRT integration to generate efficient inference code for running on NVIDIA’s GPUs, utilizing the TensorRT libraries. My contributions can be found on my github account.
Developing ops for a new MLIR dialect that helps convert MHLO / Stable HLO for NVIDIA GPUS with PJRT backend to XLA.
(Near future) Contributions to the IREE project from NVIDIA.
Mentoring IC1-IC2 engineers through multiple mentorship programs at NVIDIA.
Member of WIT Mentorship team at NVIDIA, Women In Compilers and LLVM meetup groups.

March 2018 - May 2022

Lead the testing effort in Embedded Vision and C++ Code Generation area.
Identified common problems and developed test strategies for a 50+ membered cross-functional team to utilize.
Improved performance of production code generation workflow examples in Image Segmentation, sentiment recognition (OpenMP, OpenCV C++, MATLAB and MATLAB unittest)
Created Computer Vision Simulink blocks with Multi-core C++ Code generation with compatibility and their associated examples.(MATLAB/C++)
Introduced and implemented LSTM and Attention based Neural Language Models to generate Simulink modelsthat can be used for fuzzing models during randomized testing. (Python,MATLAB / Lib:trax, Numpy, Tensorflow,stellargraph / Git, Docker, P4V)
Coached and supervised 10+ interns and new team members by developing a system for graduated learning and feedback

May 2018 - Aug 2018, Boston MA

Developed tools using dynamic search techniques for finding leanest set of component dependencies, Conversion of Hand-drawn Images to Simulink Models using GoogLe Net and Morphological Image Filters, Style Transfer between different Musical Instruments using MIDI.
Responsible for improving and maintaining the test coverage of Simulink-Component Based Modeling using MATLAB, Simulink and MATLAB Unit Testing Framework.

September, 2017 - May 2019

Enhanced the run-time of Scene Cache Algorithms by 30% using massive parallelization (CUDA/C++)
Vision and Instrumentation Lead of Backwave Technologies team. My Low-fidelity hardware prototype to recognizecoronary heart failures was selected for Venturewell accelerator's Stage II of E-team Grant Program.
Defined requirements and designed a multi-sensor system for automated guiding of a needle through Porcine spines.(Protocols/Sensors: SPI, I2C, Accelerometer, Force Sensors and Distance Sensors.) (C++, Python)

Jan 2018 – Dec 2018

Taught Object oriented programming and STL concepts in C++ to Freshmen and Sophomores.
Organized and lead 2-hour lab sections for 30+ students twice a week.
Oversaw and ran office hours for 50+ students per session to troubleshoot issues with their C++ Projects.
Prepared various mini projects in C++ on Computer vision, Game design and Machine learning for students

Jul 2015 – May 2017

Software testing experience in Embedded and Wireless Z-Wave Smart Home Automation
Developed various suites for regression and acceptance testing in Java,TestNG, and Embedded-C Framework.
Independently developed applications in Swing,Java
- A server-client system to remotely access and run test-automation suites
- A Screen-shot capturing tool to wirelessly capture the screenshots of Linux based XL*** thermostats for which I received a spot Award

Master’s in Electrical Engineering, Spec: Embedded Systems and Computer Vision

Bachelor’s in Electronics and Communications Engineering