Prawal Gangwar

◦ Implemented Machine Learning algorithms for Distributed Systems by leveraging the power of High Performance Computing
◦ Used Hadoop for distributed data management, Harp plugin for collective communication and Intel DAAL libraries for Machine Learning algorithms to get the best performance on each frontier
◦ Tested the performance against several other implementations

◦ Implemented System calls pertaining to Fork, Exec, Join, Sleep and Exit
◦ Implemented UNIX, First in First Out, Round Robin and Shortest Job First job scheduling algorithms
◦ Implemented Shared Memory, Demand Paging and Page Replacement using multiple algorithms

◦ Prototyped an IoT based smart Lock connecting to an Android App via Telegram API using QR Code for authentiation
◦ The hardware was setup using Raspberry Pi, Wifi module, PiCam and LCD Display
◦ The difficulty of authorization varied for different use cases of owner, known and unknown users

◦ Implemented an algorithm to magnify the subtle changes like change of skin color due to blood flow and small motions like that of a guitar string vibration which are not visible to the naked eye.
◦ Implementation mainly involved creating a Gaussian/Laplacian Pyramid followed by temporal bandpass filtering to reveal changes of some frequency.

◦ In automotive productions systems peripheral interfaces are very limited. For devices/ECUs (Electronic Component Units) in production firmware images need to be flashed to persistent memory. Upon booting these images are executed and ECUs start working in real car.
◦ Target was to build a Windows PC based tool that enables flashing as well as reduced flashing time on production line. This is demanded by many TI customers and is of great value of them.
◦ Building of this tool involved… Show more

◦ In automotive productions systems peripheral interfaces are very limited. For devices/ECUs (Electronic Component Units) in production firmware images need to be flashed to persistent memory. Upon booting these images are executed and ECUs start working in real car.
◦ Target was to build a Windows PC based tool that enables flashing as well as reduced flashing time on production line. This is demanded by many TI customers and is of great value of them.
◦ Building of this tool involved quick ramp-up on TI’s existing software which involved Bootloader and Software Development Kit; it also involved working with Windows libraries on PC.
◦ Successfully built a robust tool that can be used for flashing existing TI devices (TDA2xx, TDA3xx).
◦ This tool can also be extended to future devices. It supports flashing over USB and UART interfaces. Show less

◦ Implemented an Extended Kalman Filter based RGBD SLAM using Xbox Kinect Sensor.
◦ Implemented the algorithm from scratch using OpenCV and Freenect libraries on Pioneer P3-DX using ROS platform.
◦ As a part of Course Project for EE698G: Probabilistic Mobile Robotics

◦ Detected obstacles and their distance in the path of an autonomous vehicle mounted by a stereo camera
◦ Computed the free space available in front of the vehicle, by building a depth map, detecting and
removing the road, sky and background, and identifying potential obstacles
◦ Used optimization and noise removal techniques like Watershed, Erode and Dilate to segment out sky and the free road.

◦ Implemented Objective Video Quality Estimation algorithms for videos transmitted over a network
◦ Analysed several algorithms for Video Calibration,Quality Parameters and Model Development
◦ Developed a C++ code using OpenCV libraries to quantitatively rate a video quality
◦ Implemented Spatial Information filter (SI13) to extract perceptual features from a video frame
◦ Detected and quantified distortions like Temporal Delay, Spatial Shifting and Scaling, Frame Drops… Show more

◦ Implemented Objective Video Quality Estimation algorithms for videos transmitted over a network
◦ Analysed several algorithms for Video Calibration,Quality Parameters and Model Development
◦ Developed a C++ code using OpenCV libraries to quantitatively rate a video quality
◦ Implemented Spatial Information filter (SI13) to extract perceptual features from a video frame
◦ Detected and quantified distortions like Temporal Delay, Spatial Shifting and Scaling, Frame Drops, Luminance offset, Variable Frame delays and Broken Macro Blocks individually Show less