Hi everyone!

I am Gaurab and my passion in the field of Artificial Intelligence and Data Science and worked in this field. I have been working with machine learning, deep learning, computer vision and data engineering for the last 3 years. Currently, I Graduated from University of South Dakota with Masters of Science in Computer Science.

I started my carrer doing internship in data engiineering tasks in one of a prestigious fintech company in Nepal called Selcouth Technology Pvt. LTD (ASMI). Afterwards I also was promoted as AI Engineer. At ASMI, I mostly tangled with the homography, pattern matching of the surfaces in videoframe including feature point detection in planar surface including big data technologies.

I also worked on static clickable poster based projects, 2D poster injection in recorded videos and some research on in-game advertisement After working for three years, I decided to continue my studies and joined the graduate program at University of South Dakota.

• Languages and Scripts: Python, C/C++,, C#
• Frameworks and Libraries: Pandas, Scikit-learn, Numpy, Seaborn, Plotly, Scipy, Keras(with Tensorflow), Pytorch, Darknet (For YOLO), OpenCV
• IDE: Jupyter Notebook, Pycharm, VSCode, Unity
• Database: MySql, MongoDB
• VCS: Git, Github
• Collaboration: JIRA, Trello, Slack
• Methodology: Scrum
• Big Data Technology: Apache Spark, HDFS, PySpark (SQL, MLLib)
• Visualization: Microsoft PowerBi, Python Libraries (Matplotlib, Seaborn, Plotly, Dash)
• Operating System: Linux, Windows

Selcouth Technologies Pvt. Ltd (ASMI venture)

• Static Clickable Poster based Projects: Developed a video-based outfit detection system, named "Static Clickable Poster," utilizing YOLO version 4 where the datasets are extracted from different clothing sites and annotated images using labelimg across 30 different classes of male and female dresses. Increased the model accuracy from 60% to 89% increasing the datasets to 1 million.
• Core team member for recommendation system with content-based filtering and CNN for visual features extraction for an E-commerce site within the "Static Clickable Poster '' project, enhancing user experience by suggesting similar outfits based on the sophisticated outfit detection model and annotated dataset. Also, increased the performance of the recommendation system by 20% integrating the trained model in the recommendation system.
• 2D poster injection in recorded videos: Basically, it deals with injecting 2D images onto the background of the video. Achieved 30% performance gains and increased processing speed by 10% in the "2D Poster Injection in Recorded Videos" project by applying key point detection using the available libraries of OpenCV key point detector algorithm like SIFT and SURF extensively, In video frames, and implementing an efficient process to find optimal image placement locations. Most of the part of project uses Opencv. Detecting key points, features, finding the appropriate place for placing the image in video with the help of feature detection also some research with camera calibration was also made.
• Research on in-game Advertisement: Here the main idea was to place poster (2D image) advertisement in gaming environment. For this we used Unity engine and C# including the ML Library tools present in the unity game engine itself.

Selcouth Technologies Pvt. Ltd

• Wrote complex SQL scripts and PL/SQL packages, extracted data from various source tables of data warehouse and worked on Snowflake environment to remove redundancy and load real time data from various data sources into HDFS using Kafka. Also, installed, configured, and monitored Apache Airflow cluster and used AWS S3 to store large amounts of data in identical/similar repository.
• Built key business metrics, Visualizations, dashboards, reports with Tableau.
• Involved in building the ETL architecture and Source to Target mapping to load data into Data warehouse.
• Developed Spark scripts by using Python shell commands as per the requirement. Also, wrote Python for regular expression (RegEx) project in Hadoop/Hive environment for big data resources.

Applied Artificial Intelligence Lab (2AI Lab),Department of Computer Science, University of South Dakota

• Taught “CSC 586: Datamining” to a class of 55 undergrad and graduate students.
• Implemented multimodal learning techniques for chest x-ray report generation, utilizing IU datasets where only frontal images of chest Xray and its corresponding reports was taken in account.
• Developed an encoder and decoder-based multimodal framework, enabling the generation of accurate textual reports from chest x-ray images and associated text descriptions which yielded a 12% boost in BLEU score than the existing method.
• Tech: Pandas, Tensorflow, Scikit-learn, Pytorch, NLTK, NLG, Plotly, Git, Weights, and Biases.

Project Assosication For Computer and Electronice (PACE) 2018-2019
Treasurer
Computer Club, Advanced College of Engineering and Management, Kathmandu, Nepal

Graduate Teaching and Research Assistant 2022-2023
Applied Artificial Intelligence Lab, University of South Dakota

University of South Dakota
Masters in Computer Science, AI Specialization (Aug 2022 - Dec 2023) GPA: 4.00
Related Coursework: Computer Vision, Pattern recognition and machine learning, AI in Medical Imaging

Tribhuvan University
Bachelor in Computer Engineering, ACEM (Tribhuvan University) (2015-2019) Percentage: 73.77%