đź‘‹ Hi, I’m Ayushi Gupta
Computer Engineering student at Aligarh Muslim University, passionate about Backend Development and AI systems.
About Me
I am Ayushi Gupta, currently in my final year of Computer Engineering at Aligarh Muslim University. I am passionate about backend development, AI, and building scalable software systems. Over the past few years, I have gained hands on experience through internships and projects involving Spring Boot, microservices, machine learning, and deep learning. I love solving challenging problems, learning new technologies, and turning ideas into real-world applications. Outside of coding, I actively participate in competitive programming and tech communities to sharpen my skills and help others grow.
I have 6+ months of experience in the software development domain and have completed internships as:
Professional Experience
I developed a deep learning model to generate
Radio Environment Maps (REMs), which are spatial representations of wireless signal strength across an indoor area. These maps are crucial for understanding how radio signals propagate in complex environments and for planning efficient network deployments. My model not only produced highly accurate REMs but also suggested
optimal Access Point (AP) placements to maximize signal coverage, even in the presence of obstacles and varying transmitter positions using just indoor layout of the environment. With a
median prediction error of less than 3.5 dB, the system provided reliable and precise insights for indoor wireless network optimization.
Project Report : Link
I worked on building a Full-fledged prototype of an
Unmanned Ground Vehicle (UGV),a robotic vehicle designed to operate on the ground without direct human control, under the guidance of Prof. Krishna Mohan. My focus was working on
autonomous navigation of UGV, enabling it to sense its surroundings, and make navigation decisions on its own. To improve the vehicle’s decision-making, I developed a Graph Neural Network (GNN)-based adaptive path planning algorithm, which was trained on LiDAR point cloud data. This approach significantly enhanced the accuracy of navigation, achieving an 85% improvement in path planning performance compared to traditional methods.Google Colab : Link
I gained hands-on experience with Salesforce CRM, working extensively on the object model and lead lifecycle management. My role involved building process automations using Flows, Process Builder, Apex (classes and triggers), SOQL, and DML, while adhering to best practices. I also worked with Lightning App Builder, Aura Components, and Lightning Web Components (LWCs), gaining exposure to deployment strategies across environments. Additionally, I developed automation utilities and streamlined workflows using Apex (a Java-like programming language), which significantly improved lead generation processes. By optimizing database operations with SQL query aggregation, I was able to reduce system load by 40% and enhance overall performance by 60%.
Projects
I developed a
Full-stack URL shortener application that makes it easy to shorten, manage, and share links. The
backend was built with Spring Boot, the frontend with React + Vite, and the application used a PostgreSQL database for storage. To ensure secure access, I implemented
JWT authentication with Spring Security, along with features like subdomain redirects through domain aliasing and detailed URL and date based analytics displayed on a user-friendly dashboard. For deployment, I Dockerized the backend and hosted it on Render, while the frontend was deployed on a custom domain via Netlify, with the database running on a cloud-hosted PostgreSQL instance (NeonDB).
Demo: Link
Github: Link
I built a
backend microservices-based fitness tracker application using
Spring Boot, designed to track user activities and provide personalized analysis and recommendations through the
Gemini API. The system is composed of
independent services for users, activities, and AI-based recommendations, all managed via
Eureka Server and communicating seamlessly through
Spring WebClient. To handle asynchronous interactions between the activity and recommendation services, I incorporate
RabbitMQ, ensuring smooth and reliable message delivery. The application also integrates OAuth2 with PKCE using
Keycloak, providing secure authentication and authorization for all users.
Github: Link
- CNN Based Malaria Diagnosis:
I developed a
LeNet-based CNN to classify blood smear images as parasitized or uninfected,
achieving 94% accuracy. Using both TensorFlow’s functional and sequential APIs, I built custom layers and optimized key metrics such as
ROC and
specificity. I also streamlined data loading, preprocessing, and batch handling for large image datasets using TensorFlow’s Datasets API, ensuring efficient and scalable model training.
Google Colab: Link
Github: Link
I developed an
AI powered chatbot that uses
Gemini API to provide intelligent, real time responses and enhance user interactions. The application features a
Spring Boot backend for handling API requests and a React + Vite frontend for a responsive, interactive interface. Using Docker, the system is fully containerized and scalable, delivering smooth API integration, efficient backend processing, and a seamless user experience.
Demo: Link
Github: Link
Technical Skills
I love building applications, solving challenging problems, and exploring AI and modern backend technologies.
Here’s a snapshot of the skills I work with:
Langauges & Databaes:
Java
Python
C/C++
SQL
MongoDB
PostgreSQL
Tools and Frameworks:
Spring Boot
Github
Docker
Postman
RabbitMQ
Keycloak
Linux
Technologies:
RESTful APIs
Microservices
API Gateway
Eureka Server
Spring WebClient
Machine Learning (ML)
Deep Learning (DL)
Natural Language Processing (NLP)
Data Structures & Algorithms (DSA)
I enjoy connecting with like-minded people and discussing the latest in backend development, AI, and emerging technologies. Feel free to reach out to me at ayushi.gupta.9370@gmail.com.
I’d be happy to chat and share ideas!