I'm a Computer Science Graduate Student at Texas A&M University - Kingsville. with a keen interest in the exciting world of Machine Learning and Artificial Intelligence.
My passion for problem-solving and developing real-time solutions drives my work, which can be explored in further detail on this website.
My fascination with Machine Learning models motivates me to create comprehensive systems that can be utilized in today's constantly evolving world. As the Machine Learning space is still relatively unexplored, I am always eager to push boundaries and develop novel solutions.
As a hands-on engineer, I particularly enjoy constructing systems from scratch and working closely with other skilled professionals. My experience and expertise in this area enable me to develop high-quality, innovative solutions.
When I'm not in front of a computer, I enjoy a range of activities, from dancing to music to taking evening walks or spending time with friends.
- Collaborate with cross-functional teams to design and develop innovative software solutions for enterprise performance management.
- Conduct thorough testing and debugging of software to ensure high-quality deliverables.
Setup of Hyperledger Fabric on raspberry pi and implementing Modular Blockchain network (segregating fabric network, channel and peers into different raspberry pi’s) using raspberry pi’s stack.
- Worked on Shop the Look project which enables users to shop the same or similar clothes and other fashion accessories from the videos users watch on social networking platform Chingari.
- Built a Deep Learning CNN outfit colour identification model which identifies colour of clothes with 94% accuracy, and goggles colour identification model using Scikit-learn for determining goggles colour by 96% accuracy.
- Built a collar detection model trained in darknet with yolov4 architecture for identifying different types of collars of shirts and tshirts.
- Worked closely with the content moderator team and developed Flask API integrated with Spam Video Detection Yolo model for identifying spam videos from video feed. This reduced the content moderator team manual overhead by 95%.
- Built Content based News Recommendation system for suggesting similar news articles based on user’s interests and an API using Django to fetch user activity and send similar news articles as response. This system increased user engagement to the news by 20%.
- Developed an Employee Activity Monitoring software using python and tkinter for Windows and Linux which can track applications used, websites visited, keystrokes pressed by each employee along with screenshots taken at specific intervals. All the data is then uploaded to the cloud server for each predetermined slot. This helps companies to evaluate the productivity of each employee.
Collaborated with the gas station and cab services and deployed an android application "Soujanya" using Java and Google Firebase internally in an organization, where cab drivers can raise a gas filling request for a custom amount. These requests are forwarded to both gas station and cab services managers. The requests are then approved later. This helps gas stations and cab services to maintain transparent transaction records every month, clear bills on time with ease, and track cab services monthly spending on gas.
Developed a real-time web application using PHP and AngularJs to maintain complete details of one or more bank account transactions of an organization which helped the organization to monitor all transactions made from all bank accounts the organization has.
- Developed a Fingerprint Biometric System in VB.Net using SQL database in backend which includes Registration, Verification and Identification modules. These modules enable new users to Register and their encrypted fingerprint data is stored in the database and Identify and Authorise the existing user's via fingerprint.
- Created a web page that highlights various server locations along with their status on map with various colors based on the server data fetched from the database.
Developed a voice module which converts voice commands from user into text commands using google speech-to-text (STT) engine. These commands are then sent to raspberry pi which sends control signals to home appliances. Developed an android application using java integrated with Vuforia Object Scanner library to identify objects and send appropriate signals to raspberry pi to control home appliances. The assistant is easy to use and specially designed for visually challenged, speech impaired and physically disabled people.
Developed a machine learning regression model to predict the rating of an intern based on his skills in various aspects as a freelancer for a company to evaluate the performance of new interns.
The Aim of this project is to predict the price of the house, based on multiple features of the houses without using built-in Machine Learning model. Applied Gradient Discent to get the best theta values. Achieved accuracy equivalent to a built-in ML model.
Developed a regular chatting application using Java in which server can accept requests from multiple clients and clients can chat with server and among themself simultaniously.