Engineer Assistant

Abraham Jose

*** **** ***** *****, ********@*****.***

Apt 101, Oviedo Ó +1-321-***-****

FL 32765 /abraham-jose /abramjos

EDUCATION

University of Central Florida Orlando, FL, USA

Master’s in Computer Science; Emphasis on Computer Vision GPA: 3.94/4.0 August 2019 - July 2021 Relevant Coursework: Computer Vision, Algorithms & Data Structures, Data Visualization & Special Topics in CV, AR & VR, Computer Vision Systems, Advanced Computer Vision, Data Intensive Computing TKM College of Engineering, University of Kerala Thiruvananthapuram, India Bachelor of Technology, Electronics and Communication Engineering; GPA: 7.96/10.0, converted 3.36/4 Aug 2013 - April 2017 SKILLS

Languages: Python (proficient), C++ (basic) Hardware: Nvidia Jetson TX1, Rpi, Arduino

Libraries: TensorFlow, PyTorch, Unity, OpenCV, Numpy, Pandas Techs: AWS, GitHub, Docker, API, TensorRT EXPERIENCE

CRCV(Center for Research in Computer Vision) UCF, FL Research Assistant (under Prof. Abhijit Mahalanobis/for DRS Leonardo) Jan 2020 - present

Development of optimized CNN model for action recognition and detection. Optimization of the model. Graymatics Inc. Banglore, India

Algorithm Engineer (Deep Learning) Sept 2018 - July 2019

VISTA & Sherlock: Developed CNNs for scalable Analytic platforms for Smart City application using rest API & python. Traffic monitoring at the toll booth for automation - PLUS-UEM group, Malaysia(multi-model, detector & classifiers) Tracking a vehicle of interest (Color and Model, BeerNet, BCNN and Resnet, multi-class classification)

Optimization: Used Deepstream and TensorRT from Nvidia for optimization on CNN models for 3x faster inference. Sticheo Pte. Ltd. Remote

Advisor Vision Engineer Early startup Sept 2018 - December 2018

Brainstorming & bootstrapping for scalable video analytics engine for intelligent in-video ad placement.

Implemented 3D-CNN model for video boundary detection using python and tensorflow. Cell Propulsion Pvt. Ltd. SDC CV-Full Banglore, India Computer Vision Engineer Nov 2017 - Sept 2018

Founding team member of the company and lead a team of 5 as Perception team, including me.

As a team developed architecture and perception stack fusing camera and radar for Cell Prop Electric Vehicles.

Deployed models with 85% accuracy and real-time performance( 10fps) including the self-driving model.

Pitching and presentation of products and technology milestones to upper management & stakeholders. PROJECTS

Scene Cut/Boundary Detection on Videos(3D CNNs):

Implementation of Ridiculously Fast Shot Boundary Detection with Fully Convolution Neural Networks .

Achieved realtime performance and used 10 temporal frames to analyze boundary with 87% accuracy or above.

Detection of Exudation in Retinal Images:

Implemented UNet with large data augmentation(dataset-200 retina images & exudations). Dice coefficient=

.6.

Data Science Bowl 18 Kaggle Competition:

Used UNet and my augmentation library; delivered 95% accuracy in validation set for detecting cell nucleus.

The augmentation library yields 2 augmented images on a unique random seed on 7 real-life features.

AutoCar : Autonomous Rover ( Undergraduate Main Project ):

Autonomous Rover using end-to-end CNN to navigate using PioletNet on Nvidia Jetson TX1 and Rpi.

Used Haar Cascade based traffic signal detectors for demonstration. Achieved real-time(30fps) performance.

224X224 RGB images from RPi camera were sent to Jetson using python socket and trained on 4 key commands.

Wall-E : Autonomous Waste Basket ( Undergraduate Project ):

Rover to collect litter using basic image processing techniques employing Matlab, used Arduino for rover control.

Used the GUI from Matlab and pipelined Matlab to Arduino communication for controlling the rover.

Contact this candidate