Management Service Assistant

Kunal Kumar

********@***.*** 480-***-**** LinkedIn

**** * ******** ****, **, Tempe, Arizona 85281

EDUCATION

Masters’ in RAS (Artificial intelligence) Aug’19-May’21 Arizona State University (ASU), Tempe, Arizona GPA- 3.67 Bachelor’s in Information Technology July’14-June’18 Uttar Pradesh Technical University, Lucknow, India GPA- 3.6 SKILLS

Programming Languages: Python, C/C++, Java, R, Go, MATLAB Databases: Oracle SQL, MongoDB, NoSQL, Spark SQL, DynamoDB Web Technologies: JavaScript, Angular.js, Node.js, Ajax, TypeScript, React,PHP, HTML, CSS Machine Learning: ROS, OpenCV, TensorFlow, PyTorch, MLlib(Spark), PredictionIO, DialogFlow Software Platform / Development Environment: AWS, Arduino IDE, Raspberry Pi, Eclipse, Jira, Jenkins, GIT, AWS, Ant, WMA PROFESSIONAL EXPERIENCE

Graduate Research Aide Part-Time Arizona State Univ Neuro-robotics Research Lab Tempe, Arizona, USA May’20-May’21

- Programmed an Autonomous system which gather data from LIDAR sensor and camera and send the data to the cloud hub.

- Remodeled an AI and Machine Learning techniques using TensorFlow library to help maneuver, detect & avoid different obstacles

- Improved the navigational accuracy from 64.87 % to 91.88% by using Kalman filter & deep learning technique i.e., CNN Associate Professional DXC Technology Noida, India Oct’18-june’19

- Developed a python API for auto-update logs for the log management service which reduces log searches by 35-40%

- Automated product structure analysis using multi-threading and produced core health document using big data processing with ETL

- Implemented dynamic web pages based on HTML and JavaScript for client verification and validation.

- Restructured distributed system framework to process and extract request XML’s schema in NoSQL Software Developer Intern National Informatics Center New Delhi, India May’17-Aug’17

- Engineered an AI based Attendance Management System for employees in a Non-profit Organization

- Evaluated system design and business needs and technical solution which helped lowering the administrative cost by 25%

- Automated product structure analysis with multithreading and produced the core result using cloud data processing. RESEARCH PROJECTS

Trash Separating System Training system to separate trash Machine Learning (CSE 575) Jan’20-May’20

- Convolution Neural Network (CNN) was used to extract features from image and classify different types of trash

- Model was trained by gathering trained supervised data from ImageNet datasets which has 1.2 million datasets

- Data preprocessing was used for data efficiency and final accuracy of the trash separating model was 86% Navigation Movement governed by Deep Reinforced learning Artificial Intelligence (CSE 571) Aug’19-Dec’19

- Engineered the Turtle-bot movement in Gazebo Simulation using Robotics Operating System (ROS)

- Worked on Training the network and Creating, Managing and Manipulating the environment using machine learning

- Figuring hyper-parameters and facilitating Q-Value update to reflect in the robotic network GOLFIE Robot Learns to play Golf Advances in Robot Learning (CSE 591) Jan’20-May’20

- Trained the system using machine learning algorithms like Stochastic Forward Pass and Deep Learning Algorithm using PyTorch

- Integrated Computer Vision OpenCV library for image recognition to gather information about the position of the hole

- Generated the required Torque needed for the robotic hand to hit the ball so that it reaches to the hole Pursuit Evasion in Gazebo Simulation Simultaneous Localization and Mapping (SLAM) CSE 598 Jan’21-May’21

- Created a building environment in Gazebo simulation with Turtlebot3 robotic model with a camera attached on it

- Generated the map using SLAM with GMapping and provided goal location using RVIZ-GUI to move the robot

- Detecting the movement of human using HOG Algorithm and moving the robot to the human without hitting any obstacles Word-Math Problems Answering Mathematical Problems Natural Language Processing NLP (CSE 576) Aug’20-Dec’20

- Used different Machine learning models to solve Word Math’s problem and augment them for better results

- BERT and NumNet+ machine learning models were used to train on synthetically created dataset.

- Used to generate results on multiple choice questions, Paragraph questions and true/false questions. EXTRA-CURRICULAR AND LEADERSHIP

- Teaching Assistant (TA) and Grader for Machine Learning for Engineers (MAE 598) Aug’20-Dec’20

- Technology Assistant at Arizona State University responsible to secure and manage ASU servers Aug’20-May’21

- President of “Computer Forum”, a CS Students Coding Club, for 2 years during my bachelor’s degree Jan’16-Jan’18

Contact this candidate