Was a key member of the new graphics compiler frontend and midend. Involved since the inception to the delivery of the project from end to end.
Architected and implemented the Apple Metal Frontend and worked closely with Apple engineers to make it functional.
Implemented new frameworks for reading metadata coming from different api’s.
Designed and implemented the DirectX 10, 11 frontend including the parser and code generator, making it modular to be extendible for future api’s using design patterns.
Added infrastructure to be able to detect memory leaks.
Designed and Implemented the Geometry Shader and Compute Shader units of the compiler.
Prototyped memory allocators for benchmarking compile time performance improvements.
Implemented various optimizations for the compiler to improve performance leading to multiple patents.
Debugged various games, WHQL, Metal unit tests, OCL Khronos tests for both performance and functional issues.
Managed the team for a period of 3 months when manager was on sabbatical.
Mentored multiple interns and managed contractors.
Was responsible for creating and productizing multiple tools including a first of its kind shader debugger (C#, WPF), a dashboard for measuring statistics and performance data (Django) and tools for managing machine pools (C#).
Led the Bring up for Kabini Fusion APU, which got completed in a record 3.5 days with an aggressive schedule of booting Win 7 & Win 8 and running 3D Mark 11.
Led the feature, development and debug support for Kabini Fusion APU.
Involved in meeting memory requirements for driver for Windows 8
Worked on various features for improving performance for Southern Islands asic family.
Developed shader binary compression algorithms to reduce memory footprint.
Got driver feature ready for the upcoming asic products with emulators.
Have been a major stakeholder for a shared component, used across kernel mode driver, Apple OpenGL driver and linux platform.
Re-architected the graphics pipeline for the emodel© engine to achieve a significant performance improvement for heavy meshes to improve rendering and user interaction using C / C++ and OpenGL.
Implemented Core Engine enhancements in the Geometry based algorithms using C / C++.
Did UI design based on Qt and C++ for the engine.
Worked on embedding Python and Tcl into the core engine to enable embedded scripting.
Worked on geometric algorithms for enhancing the crown building algorithms for auto-building of various parts of a tooth.
Was involved in development of a metric database system using Django for the production tracking of the ICON™ digital assembly line.
Conceptualized, developed and prototyped the Metal Occlusal product for the ICON™ dental production system and eventually got it to production.
Designed and architected the emodel 9.0 software for Orthodontists.
Designed and implemented the AutoGrid application using C# which integrates the ICON system with the Rapid Prototyping system.
Facilitating dynamic runtime transformation of graphics processing commands for improved graphics performance at computing devices. (Filed 2015)
Apparatus and method for Efficient Prefix Sum Operation (Filed 2015)
Hardware Instruction Set to replace a plurality of atomic operations with a single Atomic Operation (Filed 2014)
Thesis - Surface Reconstruction from Unorganized Point Cloud Data using Incremental Delaunay Triangulation
Working on a dynamic language jit using python
Experimenting with Luajit for python jit support. Managed to get some basic constructs running.
Designed a cloth simulation using OpenGL Compute Shader support.
Rendered OpenStreetMaps and used djikstra for optimal path computation with python and Vispy.
Rendering of OpenStreetMaps using opengl on android.