Currently, I'm working at Eversource Energy as a data analyst intern to analyse sensor data from the newly modified power grid systems across Massachusetts in order to estimate system performance and quantify customer energy savings for the Grid Modernization program. I have developed time series based models with exogenous variables to statistically prove the benefits of the Volt-Var optimization algorithm by reducing the error rate (MAPE) from 6.1% to 4.19%. I also built python scripts to automate data anomaly detection that minimized manual processing of large data files. On weekly basis I lead calls with company leaders to present model results, visualizations and address issues with data quality.

During my 2022 summer holidays, I worked at Healthedge as a data analyst intern where I collaborated with the Devops team to convert client’s server configuration data into insightful dashboards. I learnt Powershell scripting language within 3 days and optimised Powershell scripts that were loading large datasets to Azure by adding filter parameters and creating pipelines. I automated the ETL process using PowerBI Gateway between Azure and PowerBI. I worked with DAX, data blending, hierarchies, cross filtering, slicers and drill through filters to create meticulous reports. These dashboards reduced the manual efforts and time for developers, which consist of client-server configurations and changes accessible all at one place in real-time.

Before coming to WPI, I worked as a Conversational Bot Engineer at Quantiphi where I created chatbots using Google's Dialogflow and Google Cloud Platform. I researched on intent detection and customizable named entity recognition approaches for chatbots in educational, healthcare and financial domains. The chatbots were also converted into voice automated calls and integrated with Big Query to store millions of conversations. The conversations were reviewed and iteratively improved which reduced the false positive rate from 35% to 3%. I mentored and trained interns in chatbot engineering on a monthly basis. Apart from building chatbots, I also worked creating a post call analysis tool for call centers. RoBERTa model was used to identify and tag sentiments detected in the call logs. This brought down the time and the cost that was required to manually analyze the huge number of calls.

Prior to this, I also worked as a Research Intern at 99Yrs Network where I partnered with the executive team to analyse data for future sales prediction, apply content-based filtering, perform customer segmentation, and deciding marketing strategies for 20+ eCommerce mutilnational companies. I built out data reporting infrastructure from ground-up using AWS for data storage, Tableau, Python and SQL to provide insights into business KPIs, marketing funnel and revenue generation