In this role, I am responsible for:<ul><li>The development of AI algorithms for driver and occupant monitoring in the car using video from in-cabin camera in real time. </li><li>Analysing limitations of the algorithms and data; subsequently advising data collection strategy.</li><li>Taking the ownership and full responsibilty of the feature I am working on. </li></ul>Features I am currently working on:<ul><li>Driver smoking and phone use behaviour detection events detection.</li><li>Real time camera pose estimation when the camera is moved by the driver.</li><li>Workload estimation of air traffic controllers from analyzing their behavior in videos in real time.</li></ul>

Senior Machine Learning Scientist

As a AI Scientist/Engineer, I was resposible for:<ul><li>Analysing the medical imaging data, documenting processes for regulatory compliances.</li><li>Collabrating with the medical doctors to gather product requirements. </li><li>Developeing ETL pipelines for the large scale imaging data for training and validation of AI Algorithms.</li><li>Developing deep learning model for diagnosing disease from CT chest images.</li></ul>

Senior AI Scientist

In this role:<ul><li>I was responsible for developing advanced image acquisition and reconstruction methods for MRI.</li><li>Collabrating with industry partners and funding agencies.</li><li>Dessiminating knowledge to the collabrators and peers by presenting work in conferences nad invited talks. </li></ul>Particularly I worked on the following projects:<ul><li>Deep learning based MR image reconstruction: Apply the concepts of MR physics and use deep learning methods for reducing the scan time in MRI acquisition.</li><li>Deep learning based motion correction in MRI: MR imaging scans are often 	lengthy (30-60 mins), and the patient movement during the scan 	causes motion artifact in MR images. Developed a new method based on deep learning to remove the motion artifacts from the images.</li><li>Brain tumor segmentation from multi-parametric MR images: Developed deep 	learning models to segment the region of the brain tumor from multi-parametric MR images.</li><li>RF pulse sequence design for accelerated imaging: Developed sequences 	for accelerated MR imaging using Siemens IDEA platform (C++ framework for MRI radio-frequency pulse sequence design)</li></ul>

Research Fellow

Tensilica’s core product for imaging is its customizable IVP (image and video) processor, which is a SIMD (single instruction multiple data) processor. I am responsible for the development of imaging algorithms using processor Intrinsic and C programming. My responsibilities and expertise at Tensilica include:Research and Implementation of imaging application:<ul><li>Responsible 	for literature survey and finding an innovative solution to imaging 	problems.</li><li>Worked 	on design and implementation of viola Jones face detection 	algorithm, pedestrian detection, and Optical Flow LK tracking.</li><li>Designed and implemented region of interest based (ROI) search for face detection in video.</li><li>Designed and implemented a skin tone based face detection algorithm.</li><li>Improved performance of face detection application to 4x using ROI, skin tone and scatter-gather.</li><li>Designed and implemented pseudo-floating-point macros.</li><li>Worked 	on implementation and optimization of Deep Machine Learning 	(Convolutional Neural Networks) for image classification on 	Tensilica IVP processor.</li></ul>Benchmarking for Customers:<ul><li>Responsible 	for design and maintenance of imaging algorithms.</li><li>Designed and implemented the solution of a system of linear equations using the Gauss Elimination method.</li><li>Worked 	on implementation and optimization of Optical Flow LK, pedestrian detection, Hough Transform, Perspective Coefficient Estimation and Matrix multiplication on the SIMD processor.</li></ul>Product Release:<ul><li>Worked 	on ISA enhancement and evaluation for new imaging processor release.</li><li>Performance analysis of code and improvement based on the analysis.</li><li>Finding and resolving bugs through bug tracking/management software (JIRA)</li><li>Documentation of software and managing the development of software through version 	control system (Perforce)</li></ul>

Lead Design Engineer

kamlesh pawar

Face Recognition Application 

Driver Monitoring System

**PhD Thesis: Compressive Sensing and Parallel MRI for Accelerated Imaging**

Magnetic resonance imaging is an imaging modality burdened by slow data acquisition process. The good
contrast between the soft tissues and imaging of arbitrary orientation slice are some of the main advantage of MR images. However long and slow data acquisition time results in patient discomfort and low resolution
images that limits the use of MRI for dynamic imaging. Much research has been done to optimize hardware for faster image acquisition but now the limit has already been reached because fast changes in magnetic
gradient causes the stimulation of the nerves in human body. So it is imperative to minimize the data
acquisition time by optimizing the pulse sequence and reconstruction algorithms on the MRI scanner. This can be achieved by designing new RF pulse sequences and new algorithms for image acquisition and
reconstruction. My PhD thesis aimed at designing novel pulse sequence and reconstruction algorithms using
compressive sensing (CS) and parallel imaging (PMRI) techniques to address problem of slow data acquisition
in dynamic MRI. 

**Master Degree Project: Embedded Optical Waveguide Biosensor**

This project was concerned with the development of low cost biosensor for onsite detection of protein and
pathogen based on change in optical properties of optical waveguides. The sensor was designed and simulated using OPTI-BPM software (tool for simulation of optical waveguide). The design was also realized by fabricating it using polymer SU8 and PMMA. Antigens were immobilized (trapped) on the surface of SU8 polymer, these antigens when react with analyte causes change in optical property of waveguide. The change
in optical property was measured that provides an indication on the presence of antibody in analyte

Personal details

kamlesh P.

Summary

Technical skills

Work Experience

Education

Personal Projects