Sep 2024 – Jan 2025 · 4 mos

• Client Location: Bengaluru
• Implementation (testing & validation) of libflame library tests: hetrf_rook, hetrf, sytrf_rook
• Enhancement (mostly validation improvement) of libflame library tests: hseqr, geqp3, hgeqz, gghrd
• Run-time optimisation of single thread geqrf test (by varying Block Size for various input size ranges): Up to 15% on Genoa and up to 28% on Turin for various data types

Feb 2024 – Aug 2024 · 6 mos

• Client Location: Bengaluru
• Implementation (testing & validation) of libflame library test: larf
• Enhancement (mostly validation improvement) of libflame library tests: stevd, stedc, steqr

Oct 2022 – Feb 2024 · 1 yr 4 mos · Faridabad, Haryana

• Learnt developing android apps (like Notes app, BMI calculator app, Tic Tac Toe game app, Calculator app, Simple Music Player app) and features

Feb 2022 – Oct 2022 · 8 mos · Bengaluru, Karnataka, India · Remote

• Worked on optimizing native virtual background (NVBG) module operations using Intel and AMD GPUs:-
• BGRA image (32 bits per pixel) to NV12 (12 bits per pixel) color conversion using Direct3D12 / Direct3D11 video acceleration.
• Computed mask and alpha using Direct3D12 compute shader. Obtained CPU utilization improvement of 17%.
• Enabled compilation of NVBG module for Windows Arm64 - Included compilation of source code & dependencies libyuv, DirectXTK12

Aug 2021 – Nov 2021 · 3 mos · Bengaluru, Karnataka, India · Remote

• Worked on learning about generation of the various deliverables of the solar product, and fixing issues.

Jan 2019 – Aug 2021 · 2 yrs 7 mos · Bengaluru, Karnataka

• Took time off for personal well-being and self-improvement. Used the period to reflect/build resilience/upskill before rejoining the workforce.

Aug 2017 – Jan 2019 · 1 yr 5 mos · Bengaluru, Karnataka, India · On-site

• Division: Technology
• Migrated the generation of mini adjustments report from legacy to newer SecDB platform and set up its jobs on the automation infrastructure for daily run.
• Resolved numerous (almost all of the 30 odd provided) securitised products regression tests issues.
• Operated tech support that included morning/weekend checkouts for various jobs and solving issues with them and resolving support requests and queries on a weekly basis being part of rotation cycle.

Nov 2016 – Jul 2017 · 8 mos · Hyderabad, Telangana, India · On-site

• Provided support for OpenGL ES on MSM8998 chip, generated benchmark numbers with enabling and disabling Snapdragon LLVM compiler.

May 2015 – Jul 2015 · 2 mos · Atlanta, Georgia, United States · On-site

• Teaching Assistant to Prof. Dr. Irfan Essa (Distinguished Professor in the School of Interactive Computing and Senior Associate Dean in College of Computing) for online Computational Photography class in Summer 2015 semester. Helped students and other teaching assistants with their questions on assignments and course material on the Piazza online forum, created a schedule for grading and graded assignments.

Jan 2015 – May 2015 · 4 mos · Atlanta, Georgia, United States · On-site

• Teaching Assistant to Prof. Dr. Irfan Essa (Distinguished Professor in the School of Interactive Computing and Senior Associate Dean in College of Computing) for online Computational Photography class in Spring 2015 semester. Graded assignments.

Aug 2011 – Jun 2014 · 2 yrs 10 mos · Bengaluru, Karnataka, India · On-site

• Enabled cross-compilation of CUDA driver, apps, runtime and libraries for mobile (Android and Linux for Tegra platforms) for NVIDIA’s Logan system on chip (SoC) and fixed several issues in them.
• Developed bash scripts for CUDA bringup on pre-silicon platform (emulation).
• Implemented GPU apps using OpenCL/CUDA - Floyd Steinberg Dithering in OpenCL (~1.5x speedup), CUDA-Direct3D9 texture interop.
• Enabled interoperability of CUDA and OpenGL for mobile.
• Benchmarked CUDA performance on Logan against OpenCL performance on other vendor SoCs using CUDA and OpenCL implementations for various basic general-purpose scientific and engineering samples/applications.

May 2010 – Jul 2010 · 2 mos · Osnabrück Area, Germany · On-site

• Received DAAD – WISE Award/scholarship for this internship
• Project: Parallel Processing of Robot Sensor Data, Advisor: Dr. Joachim Hertzberg
• The background of the project is that SLAM6D is a scan registration software that is used for reconstructing maps for mobile robots from processing and interpreting data from 3D laser scans. My work involved implementing in GPU parallel CUDA language the algorithm of nearest neighbor search in clouds of 3D laser scan points to speed up the scan registration process.