Vansh Patel

• Co-founded a club for competitive programming at Arizona State University.
• Won 15 Model United Nations Competitions.
• Amongst Top 15 ranked students in undergraduate studies.
• Led and won 2 lab simulation competitions.

Python - Flask, Machine Learning, Selenium Automation and Testing

C# - ASP.NET, MVC5, .NETCORE

Javascript - Node.js, Angular, Express, React

Database Systems - MSSQL, MongoDB, MySQL, Postgres

Blockchain - Solidity, Smart Contracts, Hyperledger Fabric

Virtualization Technologies - Docker, Microservices Deployment

Domain - Web-based software engineering, API design knowledge, Database management software, Testing and maintenance, Application development, REST Development

Secure Screen Session Management System

• Devised a production-level system for secure session management without the use of 95% of expensive hardware.
• Integrated the spoof prevention algorithm used as at the core of the application which increases the efficiency of software by 50% compared to others in field based on research.
• Coded a highly secure server-based service for analyzing user data and developed a fail-safe for algorithm on the scenario of internet disconnection.
• Implemented the application in Python, OpenCV, and Flask

Audio Spoof Detection using Spectrogram

• Designed a solution for the ASV Spoof detection competition using unconventional machine learning approaches to prevent audio spoof false positives by 65% more accuracy compared to similar state-of-the-art models.
• Implemented a custom GAN model to augment data and strengthen an ensemble of 5 image recognition models developed with help of mentor and team.
• Accelerated the detection process up to 30% by enhancing the state-of-the-art algorithms to suit the detection task.

Electrical Dashboard

• Built a scalable heavy machinery parameter tracking web application using Angular, MongoDB, and .netcore
• Designed a responsive dashboard with 9 real-time graphs for 5 custom parameters, using a private server that connects to up to theoretical 300 real-time sensors simultaneously.
• Engineered features including real-time anomaly detection of machinery, 9 data graph plots, and a two-factor authentication login.

Image recognition as a service (SaaS)

• A responsive scalable image recognition service using AWS SQS, EC2 and S3 and custom load balancer logic. The frontend is built in Angular and backend is made in NodeJS.