Shivrinith Prasad

• Consistently met project deadlines by effectively managing time and prioritizing tasks according to importance.
• Analyzed proposed technical solutions based on customer requirements.
• Developed scalable and maintainable code, ensuring long-term stability of the software.
• Collaborated with management, internal and development partners regarding software application design status and project progress.

The first project that I joined was the Model Mart initiative. At Model Mart, I developed an internal application to manage all ML models at Home Depot, allowing employees to easily discover and access relevant models. In the Sorting and Ranking space, I built the L1 ranker model, which significantly improved search relevance. I also maintained its infrastructure and led KTLO efforts, including migrating CI/CD pipelines from TeamCity to GitHub Actions and moving Orange Tree jobs to Airflow.

Later, I helped evolve the L1 model into the L2 Multi-Signal Ranker. I developed the automated pipeline for this model, collaborated with the Data Science team to deploy it to Vertex AI, and integrated it with our internal APIs.

Throughout these projects, I’ve been deeply involved in the full software development lifecycle and have successfully deployed multiple solutions to production. These experiences have shaped my growth as an engineer, and I’m truly grateful for the opportunities and learning I’ve had at The Home Depot over the past three years.

This intern project was centered around the Cart and Checkout space, where I contributed to the “Checkout Anomaly Detection” project. For this project, we used Jupiter notebooks to create a python application to conduct Exploratory Data Analysis that would analyze the checkout data to detect any anomalies within the data and draw graphs based on our analysis. We used machine learning models to achieve this, both unsupervised (K-Means clustering) and supervised (K-Nearest Neighbors).

This project involved creating an internal application for the mobile team. This application would help the admin of the office better assign seats within the office. In order to ensure that this app would run on both iOS and Android, we used Flutter in order to achieve this. This intern project was mainly in place to test the functionality of Flutter and the viability for it as a replacement to the current app structures in place. Flutter was too simple and single-threaded at the time of app development, making our current implementation better.

For this project, I was tasked with using machine learning models to analyze the cart and checkout data. Differing from the other internship, this project was working with decision tree's and conducting an analysis on cart and checkout data. Through this, we were able to generate confusion matrices that aided in our understanding of the customer data.