Projects
ProColab
A collaboration platform for students of various universities to work on projects together. The platform will leverage their skills and abilities to help them find suitable partners using AWS Sagemaker to help them on their projects. We provide the users with the option to upload their resume and parse it using NLTK library. We store the user details in AWS DynamoDB and the users resume in AWS S3 buckets.
Smart Photo Album
The photo album is a scalable web application designed and deployed using AWS services like S3, Lex and ElasticSearch. CI/CD DevOps pipeline is used to automate the build, test and deploy phases. The application allows users to make search requests, display the search results resulting from the query as well as upload new photos. The user is also given a choice to use either voice or text to perform the search queries.
AIHW Benchmarking
We implement and benchmark standard Deep Learning algorithms and do performance profiling with different setting such as Single or Multi in a traditional training and distributed training, and using different configurations for the Analog memory devices using the IBM AIHWKIT. We try to benchmark the performances of the models using a single node and multiple nodes as well. We also try to benchmark the performance of the models using different configurations of the Analog memory devices. This enables us to compare the performance of the models using different configurations of the Analog memory devices. We try to write consistent code with a high level of abstraction and modularity to make it easy to use as well as to make it easy to extend.
Customer Churn Prediction
In this project I have implemented a customer churn prediction application, that can predict if a customer will churn or not. The model is based on the customer's past behavior and the current customer's behavior. The model is trained on a dataset of 100k customers. The model is tested on a dataset of 10k customers. The model is deployed on Heroku.
