Harsh currently works in the Marketing and Customer Analytics domain with the CE Finance team at Amazon
InSight - Retail platform for Facial Recognition & Analysis
This cost-effective and resilient facial recognition tool built using OpenCV & AWS Rekognition can identify customers with only 1 image used for training the tool, as well as analyse their facial patterns to extract demographic information and sentiment. This solution could be widely applied to various industries including security and retail.
Winning MinneMUDAC Data Science Challenge
The MinneAnalytics 2019 challenge involved forecasting Soybean futures so as to enable farmers to make informed decisions about selling their crop. A decision on what price to sell at, especially in the volatile market of 2019, is important. Our work involved collecting data including but not limited to commodity prices, financial indexes, google news trends, and tweets of policy makers. This was followed by an extensive implementation of predictive modeling methods including ensemble methods and recurrent neural networks. Our model achieved a prediction error of ~5.6 cents (< 1%).
More than 100 teams presented at the Optum Technology Center in Minneapolis on November 9th and our team was awarded first place in the Graduate Student division.
Automated Image Rating Tool
This work-in-progress project brings together my two passions - Machine Learning and Photography. My aim is to build a model to help professional photographers optimize their workflow by selecting the best images from a job using artificial intelligence to segregate images based on features like sharpness, emotions, facial recognition and aesthetics using Computer Vision, Deep Learning and Predictive Modeling. Below is a sneak peak from the project!
Natural Language Database Querying
Python web-application that leverages NLTK to enable business users to query their database in Natural Language (Speech or Text) and get Interactive Visualizations.
Vision behind this project: Managers today need to make informed decisions in real time. Top companies use data-driven insights to make business decisions, and it is crucial to speed-en up the process for managers and non-technical users to be able to fetch results from their database quickly and efficiently.
So we asked ourselves, what is the fastest and most efficient way to fetch results? Google, of course! Setting Google Search as the benchmark, our team of aspiring Business Analysts and Data Scientists set out to create an application where non-technical users can query their databases without any prior technical knowledge or intervention by a data analyst.
So now, why SQL? When you can Google