Harshit Agarwal

I'm a Post Graduate student at IIIT Bengaluru. Through my coursework and projects, I have gained practical experience in cleaning and analyzing datasets, applying statistical and analytical techniques to uncover patterns, and presenting insights in a clear and effective way. Previously, I was responsible for managing and running my family business, where I handled client negotiations, built strong relationships, and ensured smooth day-to-day operations to manage and grow the business.

Email  /  CV  /  Github  /  Leetcode  /  Kaggle  

IIIT Bangalore
PGD in DPDM
Jun. 2025 - Jul. 2026

MinkWhite
Ruby on Rails Developer
Aug. 2023 - Jan. 2024

BMSIT Bengaluru
B.E. in Information Science
Aug. 2019 - Jun. 2023

Visualizations

All visualizations made using Tableau.

This line chart based Dashboard visualizes profit per passenger for major U.S. airlines over time, highlighting how airline profitability evolved before and after the COVID-19 pandemic. By tracking yearly performance, the chart clearly shows the sharp financial disruption during the pandemic period and have a continued effect on the profits.

This is workbook of Tableau, visualizing the dataset of Electric Vehicle Population Data. It was designed by keeping various KPI Requirements such as Average Electric Range, Total BEV Vechicles Relative to Total Vehicles, Total PHEV Vehicles Relative to Total vehicles etc. Charts like Line/Area chart, Map Chart, Pie/Donut Chart, Bar Chart etc were used in Visualization and finally a Dashboard was made for Data Discovery.

This is workbook of Tableau, visualizing the dataset of WEF Youth Report. Dataset Link can be found Here.

This visualization addresses the challenge of enabling users to dynamically change the dimension that drives color across all charts in a dashboard. Because the available dimensions may evolve over time, the solution is designed to be flexible and easy to maintain. The focus is on creating an approach that ensures consistent color behavior, simplifies future updates, and enhances overall dashboard usability.

Data Analysis

Projects based on analysing Datasets, mostly comprising of EDA, Feature Engineering processes for the Data.

Built a XGBoost Regression Model to that helps establish the rates of premium by predicting the charges or payouts done by the firm. Achieved a total of 15-20% improvement in RMSE over baseline models such as Linear Regression. The approach for the project was
1.) Exploratory Data Analysis(EDA)
2.) Build and Evaluate baseline linear model
3.) Improve on the baseline linear model with Data Preprocessing.
4.) Improve the model training process using Sklearn's Pipeline and compare the results of final model using RMSE Error Values.
Other processes involed in the stage of developing this project was Understanding Correlation between the Categorical and Target Variables and various methods used in Correlation Analysis. Implementing BayesSearchCV for XGBoost Hyperparameter Optimization

This project analyse the Uber dataset provided in Kaggle.
The following KPI's were investigated for the best outcome of the Dataset -
1.) ARPU - Average Revenue Per User
2.) Usage Frequency per Month
3.) Monthly Active Users (MAU)
4.) Retention Rate and etc...

This project analyse the Titanic - Machine Learning from Disaster dataset provided in Kaggle.

Hobbies and Interests

Implementing Quantum Ciruits

Problem Description - In e-commerce, determining the optimal price point for products is crucial for marketplace success and customer satisfaction. Develop an ML solution that analyzes product details and predict the price of the product.
(The relationship between product attributes and pricing is complex - with factors like brand, specifications, product quantity directly influence pricing.)

We used TF-IDF Vectorization for text data and LightGBM Regression on numerical features extracted from the catalog content.

Problem Description - In this competition, you are invited to build a model that predicts the next market state from a sequence of prior states. This is a very challenging endeavor due to the complexity of the data: many standard time-series statistical assumptions are not met here. Yet the problem is feasible — Wunder Fund's 10 years of successful trading prove it. The task mirrors problems quantitative researchers face daily. You'll have to be smart — in the HFT domain, inference needs to be made under very tight time constraints, so any practical solution must be nimble enough to run on CPU.

Under Construction

Under Construction

Source code from Jon Barron