Jaydeep Joshi

MSC. COMPUTER SCIENCE

CGPA: 8.54/10

Web Developer

• Developed responsive web applications using React JS, HTML, CSS, and Bootstrap.
• Contributed to the development of e-commerce marketplaces and payment gateways for WordPress applications.
• Web pages, plugins, and functionality were designed, implemented, and monitored for continuous improvement using Elementor page builder.
• I was responsible for designing, developing, and redesigning their primary services webpage. For best usage, I produced beautiful website layouts, functionality, user-friendly design, and straightforward navigation.
• Many data-related concerns were resolved, including data duplication and missing data, which benefited in improving data quality.
• Using client needs, I developed precise specifications for project plans.
• Issues were discussed with team members in order to give solutions and apply best practices.
• Designed a front-end user interface using Figma.
• Worked on a project that was already ongoing and fixed issues.
• Improved functionality by updating old code to new development standards.

Software Developer

• Worked on building responsive and multi-platform mobile applications using React Native.
• Developed a chat board using a drift and built multiple theme functionality for IOS devices.
• Involved in testing API calls (Get, Post, Put) developed for the web portal.
• Used GitHub as a version control tool.
• Designed, implemented, and monitored web pages, plugins, and functionality for continuous improvement.
• Incorporated graphic design knowledge into web app development to enhance the design.
• Customer needs were translated into technical site concepts for preliminary planning.
• Ensured that deadlines were achieved successfully.

Web-design Intern

• Learned HTML tags, Css styles and JavaScript.
• Created a personal portfolio webpage using audio-video tags, inline css and APIs.

With a rapid rate of change in phishing attacks, the cybersecurity domain is challenged to identify a wide range of phishing attempts reliably and efficiently. Here, we focus on examining the impact of changes in feature space over the last decade on phishing detection. Three popular datasets from 2012, 2015, and 2020 were evaluated based on the similarity and uniqueness of features between them. We identified the crucial features that provide comparative results across various phishing datasets. Our experiments achieved classification results comparable to those of existing work with an average of 50% fewer features. Consequently, the detection performance of machine learning models improved significantly with minimal accuracy loss. We also achieved reasonable accuracies using ensemble feature selection algorithms, with an average of 80% fewer original features. In addition to quantitative improvements in the execution time and accuracy of phishing detection, this work helps identify the characteristic features that must be preserved across data sets to help organizations adapt to the massive growth in phishing attacks.

• In response to a request for a video clip, HOG applied edge detection, perspective transformation, and feature extraction.
• Model created using a linear SVM in machine learning with an accuracy score of 0.99.

• As project leader for this senior project, I directed a group of four people.
• Developed a web platform to evaluate user health data and provide information on nearby home remedies and medical services.

• FLASK was used to create a local host web application that allows users to post any remark as a review or comment and have the model respond with a positive or negative reaction.
• Machine Learning's logistic regression model was used, and the Accuracy Score was 0.87.

• Image-preprocessing was used to check and fix corrupted pictures, while data augmentation, dropout, and early stopping were implemented to address the overfitting and underfitting problems.
• Model created using 5 CNN layers of deep learning with an accuracy score of 0.85.

• The temperature of several parts of a permanent magnet synchronous motor is predicted using machine learning and deep learning-based models using time series data provided by Paderborn University.
• To estimate the temperature, a number of deep learning and machine learning models were used, such as CNN, LSTM, K-Nearest Neighbor, and linear regression.

• A 10-liter bucket, a CD, a toy-car motor, wires, a 9-volt battery, and newspaper were used as waste materials to create a simple, reliable vacuum cleaner for students.