Jae Hyuk Choi

adlp0c@r.postjobfree.com • 312-***-**** • github.com/pnut2357 • pnut2357.github.io • linkedin.com/in/jaechoi2357 EDUCATION

M.S., Robotics & Autonomous Systems Dec. 2020

Arizona State University Tempe, AZ Concentration: Artificial Intelligence & Computer Science GPA 4.00/4.00 B.S., Mechanical Engineering B.A., Physics May 2015 University of Evansville Evansville, IN GPA 3.87/4.00 SKILLS

• Programming: Python, C++, Java

• Databases: PostgreSQL, MySQL Big Data: Hadoop, Spark

• AI/ ML: Scikit-Learn, Tensorflow, Pytorch, Keras, OpenCV

• Design & Applications: MATLAB, ROS, AWS, Google Colab, Jupyter, LabVIEW

• Systems: Linux, Unix, Mac OS, Windows

PROJECT EXPERIENCE

Recommender System in Amazon Game Data Link Skills: Python, Surprise, OOP, Keras Feb. 2021

• Designed the best recommender system on Amazon json data to enhance E-commerce sales in converting web-site browsers into paying customers, improving cross-selling opportunity, and establishing customer loyalty.

• Coded to extract important features and perform Collaborative Filtering and Neural Matrix Factorization to compare.

• Determined SVD++ or N-MF3 for targeting users with their behavior patterns and KNN for the highest HR and cHR. Question-Answering in Medical Domain, ASU Link Skills: Python, Pytorch, OOP, Transformers April 2020

• Identified the question-answering task by developing a learning model to answer gap-filling queries.

• Coded to transform the task into a classification task by NER tagging, using collate function to pass batches of passage-query pair to the training loop of BERT so that it generates the embedding for collated pair. These embeddings are passed to CNN to classify the tag of every token by updating weights over the tokens while back-propagating. Accumulating loss to transfer to capture a better contextual understanding to answer the queries in the medical field.

• Developed a model that achieved 11% better performance of the old model (45% of F1-Score; human novice level). Hot spot Analysis on NY Taxi Trip, ASU Link Skills: Scala, AWS, Hadoop, Spark, PostgreSQL, MySQL Mar. 2020

• Coded in Scala to perform spatial queries to identify most significant hot spots based on number of taxi customers’ pick-up locations. Used Hadoop to distribute tasks to multiple (3) worker nodes under one master node and used Spark to implement data parallelism and fault tolerance.

• Identified the fifty most significant hot zones from the New York taxicab dataset. Big Bro; facial recognition intruder alert system, UOA Link Skills: Python, AWS, OpenCV2, Boto3 Jan. 2020

• Brainstormed the idea of detecting the presence of people via face detection while monitoring inside of the structure to alarm if a stranger whose photo image is not in the database.

• Coded to capture an image using OpenCV2 and to upload it as the detected image to S3 bucket using boto3. Utilized Lambda and Rekognition to compare the uploaded image with the repository of the family images stored in S3 bucket. Implemented SNS to trigger email alert to the user if detected face does not match with the family members.

• Awarded 1st Place demonstrating the application for Amazon and State Farm challenges. Data Analysis and Best Model for Heart Disease Prediction, ASU Link Skills: Python, Scikit-Learn Sept. 2019

• Defined an objective of finding the significant features and the best model for further investigation of heart disease.

• Coded to combine and analyze data to determine the features highly correlated with the heart disease, reduce dimensions with these features, and compare various models for the best one using bagging, boosting, and stacking.

• Determined 6 features to be instigated by medical doctors and Gradient Boosting to be the best (87% test accuracy and 95% combined accuracy) based on highest AUC among perceptron, Logistic Regression, SVM with linear and rbf, Decision Tree, Random Forest, and K-Nearest Neighbor.

PROFESSIONAL EXPERIENCE

Computer Vision Engineer Intern, Sensagrate Skills: Python, Tensorflow June 2019 – Aug. 2019

• Collected a point-cloud of 3-D data with intensity values from various types of LiDAR sensors and set the environment on Jetson TX2.

• Collaborated building the Python-based deep-learning model using Tensorflow for object detection and tracking on LiDAR datasets to achieve 71% accuracy.

HONORS AND AWARDS

• 1st Place for AWS and State Farm challenges at Hack AZ Hackathon Jan 2020

• Engineering Graduate Fellowship (ASU) Fall 2019 – May 2020

• 1st Place at Math, Engineering, and Science Undergraduate Research Conference (MESCON) Fall 2014

• Academic Scholarship (UE) Fall 2010 – May 2015

Contact this candidate