Business Analytics Grad Student

Programming and Analysis Tools : Python, R, SQL, Pandas, NumPy, Statsmodels, Scikit-learn, PyTorch, TensorFlow

Data Management and Visualization : MySQL, PostgreSQL, Timescale, MongoDB, Elastisearch, Spark, Kafka, Grafana, Tableau

Infrastructure and Operations: Docker, Airflow, AWS, Azure, GCP, Databricks, Git, Unix (Shell), Jira, MLFlow

Education

• M.S., Business Analytics

  University of California, Davis (June 2024)

• B.Tech., Mechanical Engineering

  SASTRA Deemed University (Aug 2019)

Work Experience

Data Scientist @ Practicum Project at a Leading B2B Lender, USA (Sept 2023 - Present)

• Analyzed and synthesized over 40 million Dun & Bradstreet records with internal datasets, using Python to uncover key trends and insights that informed strategic decisions
• Applied feature engineering to transform raw data enhancing direct marketing and customer conversion rates
• Conducted comprehensive data analysis and model creation to cut client reliance on third-party loan channels by 10% through incorporation of ensemble unsupervised model like XGBoosts

Data Science Engineer @ Vunet Systems, India (Oct 2020 - June 2023)

• Led the ETL process for top Indian banks using TimescaleDB and Docker, integrating Airflow for automation and Grafana dashboards to enable comprehensive oversight
• Leveraged TimescaleDB with efficient compression strategies, achieving a tenfold decrease in data retrieval time and reducing storage needs by 80%
• Designed intricate ML pipelines with Airflow DAGs, establishing rules to streamline workflow execution and enhance efficiency
• Designed real-time data visualization dashboards for Unified Payment Interface platforms, interpreting billions of transactions monthly, which slashed client system downtime by 10%
• Partnered with business and customer service teams to track critical KPIs, devising a statistical alert system that cut down non-essential alerts by 20%
• Played a pivotal role in enhancing our vuRCABot machine learning tool, aiding cross-functional team efforts that cut down on time to identify incident root causes by 75% for site reliability engineers
• Implemented pipelines with Kafka topics and ksqlDB streams to effortlessly transfer data from any collection agent to relational databases using Kafka sink

Research Intern @ Deakin University, Australia (Jan 2019 - July 2019)

• Employed data analytics methodologies to develop a mathematical model using an Evolutionary Algorithm, focusing on optimizing water coolant channels on a tool based on existing design data.
• Conducted finite element analysis to analyze temperature variations on an AA7075 Aluminium alloy showcasing the application of data-driven insights in engineering simulations.

Projects

Cardboard Field Experiment

This study explores that with COVID-19, our attitudes toward physical contact could have shifted, affecting our high-five choices. Our experiment shows how small visual cues can shape social behavior, with air high fives remaining popular amid ongoing concerns about health and safety. Also it explores if the gender of someone holding a cardboard sign influences people’s preference for physical or air high fives. On high level, we used block experimental design, performed hypothesis and contingency tests to explore the significance of physical interactions which could further be related and significant to business, and other promotional activities.

English Premier League Prediction Analytics

This project means a lot to me. When I was watching a Liverpool game in New York, I started thinking about how well fans could guess the outcome. I wanted to look into this further by checking how fans felt before big football matches and using that to guess who would win. Instead of using old player or match stats, I believe I could use the fans sentiments which I consider to be super fun instrumental variable. So the idea is to develop sentiment models using NLP techniques, using data from public soccer forums to predict match outcomes based on online sentiment polarities, which has the potential to help in making informed betting decisions by capturing the latest trends before the game.

San Francisco Vision Zero

Finalist in the hackathon organized by City and County of San Francisco, analyzed San Francisco traffic accidents data and engineered features for a logistic regression model, enhancing Vision Zero initiatives with actionable, data-driven policy recommendations.

You can look for other projects in my github :)

Social

Connect with me!

Github

LinkedIn

Medium