Powered by
Codebasics Bootcamp

Ayush Kishor Agrawal

Data Scientist

+91 7620063073 [email protected]

Resume

Hello, I am

Ayush Kishor Agrawal.

2

Machine Learning Projects

About Me

Hey there! I'm Ayush, a student at IIT Kharagpur — And I like to turn messy data into meaningful stories and smart decisions.

With experience in both data analyst and data scientist roles, I don’t just explore data—I build predictive models that drive real business impact.

🛠 Tools I Trust:

🐍 Python, Scikit-learn
🧮 SQL, Excel
🧹 Pandas, NumPy

🔍 What I Do:

📊 Exploratory Data Analysis (EDA)
🧠 Machine Learning & Predictive Modeling
🧼 Data Cleaning & Feature Engineering
🧪 Model Evaluation & Tuning

Always eager to connect with like-minded folks—whether it’s for project collaborations, idea sharing, or just a good data chat.
Let’s build something smart together! 🤝
Until then... dive into some of my favorite projects below 👇📊

Key Skills

Python

SQL

Pandas

Numpy

Statistics

Machine Learning

Matplotlib

My Projects

My Experience

AI/ML Intern – Infosys Springboard (remote)
Oct 2024 – Dec 2024

Developed an AI fraud detection system using YOLOv8/YOLOv9 for object detection and EfficientNet/ResNet for document classification.
Integrated OCR (Tesseract/Google Vision API) for data extraction, enabling accurate identification of mismatched personal information.
Implemented intelligent matching algorithms (Levenshtein Distance, field-level comparisons) to detect mismatched identity information
Validated key fields like name, UID, DOB, address, and phone number to enhance verification and reduce fraudulent entries by 40%.

Deployed the model through the interactive Streamlit app, enabling stakeholders to explore predictions and insights in real time through an intuitive and user-friendly interface.
Link of the Streamlit app - App link.

Awards & Certificate

Python: Beginner to Advanced For Data Professionals

SQL for Data Science

Math and Statistics For AI, Data Science

Master Machine Learning for Data Science & AI: Beginner to Advanced

Intern Certificate @ Infosys-Springboard

Intern Certificate @ Infosys-Springboard

Let's Connect

Feel free to get in touch with me. I am always open to discussing new projects, creative ideas or opportunities to be part of your visions.

[email protected]

https://www.linkedin.com/in/ayush-agrawal-061030224/

+ 91 7620063073

https://github.com/ayushagrawal4

Machine Learning Project

Healthcare Premium Prediction | ML Project (Regression)

Domain/Function: Healthcare

Developed a regression model to predict healthcare insurance premiums using customer profiles. Achieved 98% R² accuracy with Linear Regression and XGBoost. Deployed a real-time Streamlit app after data cleaning, EDA, and model tuning.

View Project
on GitHUB

View Project
on

Healthcare Premium Prediction | ML Project (Regression)

Feature importance analysis revealed that insurance plans, genetic risks, and risk scores have the highest impact on predicting insurance costs.

Healthcare Premium Prediction | ML Project (Regression)

At least 95% of prediction errors have a percentage difference of less than 10% between the predicted and actual values.

Project Details

Objective: Predicted insurance premium amounts using regression models, based on customer demographics and policy details, to help insurers make data-driven pricing decisions.

Data Prep: Cleaned and preprocessed historical insurance data with Pandas and NumPy. Conducted exploratory data analysis (EDA) to uncover feature correlations and trends affecting premium values.

Modeling: Trained and compared Linear Regression and XGBoost models. Applied hyperparameter tuning techniques to enhance model performance and ensure reliable, generalizable results.

Deployment: Achieved a 98% R² score. Built and deployed a Streamlit web app for real-time premium prediction, allowing users to interactively explore model outcomes.

Machine Learning Project

Credit Risk Modelling | ML Project (Classification)

Domain/Function: Finance

Built a credit risk classification model to minimize loan defaults using XGBoost and Logistic Regression. Applied SMOTE for class imbalance and deployed a real-time prediction tool via Streamlit. Achieved 94% accuracy and 96% recall.

View Project
on GitHUB

View Project
on

Credit Risk Modelling | ML Project (Classification)

Picture depicting AOC curve

Credit Risk Modelling | ML Project (Classification)

Graphs showing feature distribution of default and non-default.

Project Details

Objective: Built a classification model to assess credit risk and identify high-risk loan applicants, helping reduce default rates and support better lending decisions.

Data & Modeling: Preprocessed credit data with scaling and encoding techniques. Trained Logistic Regression and XGBoost models, incorporating SMOTE to address class imbalance for fairer predictions.

Evaluation: Focused on recall, precision, and accuracy as key metrics. Logistic Regression achieved 94% accuracy and 96% recall, outperforming other models in capturing high-risk cases.

Deployment: Deployed the model using Streamlit, enabling real-time credit risk forecasting through an interactive, stakeholder-friendly web interface.