Turning data into useful decisions.
I’m Kundan Pandey, a final-year Computer Science student and Data Science Intern building practical machine-learning, analytics, and computer-vision applications.
Practical AI, not just notebooks.
My work connects data preparation, feature engineering, model evaluation, and clear interfaces. I aim to create applications that make an analysis or prediction usable for a real person.
End-to-end prediction workflows, analytics dashboards, and real-time computer-vision experiences.
Machine-learning deployment, robust model evaluation, feature engineering, and stronger SQL analytics.
Tools behind the work.
01Machine Learning
02Data and Analytics
03Applied AI
04Development
Learning through applied work.
Data Science Intern - Axlero Solutions
Analyze datasets with Python and SQL, support predictive-model development and evaluation, and communicate findings through clear reports and visualizations.
Bachelor of Technology in Computer Science and Engineering
CGPA: 8.03/10. Focused on software development, data science, and applied machine learning.
Machine Learning and Data Science Training
Hands-on training in data cleaning, exploratory data analysis, regression, and classification.
From model workflow to usable application.
Use Streamlit to turn ML workflows into accessible dashboard and prediction experiences.
Deploy portfolio and frontend work through Vercel, with Render-ready configuration for Python web services.
Projects built around real data problems.
Customer Churn Analytics
Churn prediction and analytics dashboard built around the Telco Customer Churn dataset.
- SQL segmentation, churn-rate analysis, and revenue-loss estimation
- Dataset upload, high-risk identification, and prediction export
Credit Card Approval
End-to-end approval classifier using the UCI Credit Approval dataset.
- Unified preprocessing for missing, categorical, and numerical data
- Cross-validated Logistic Regression with GridSearchCV and class balancing
Wine Quality Prediction
ML web application that predicts wine quality from physicochemical features.
- EDA, preprocessing, visualizations, and feature-importance analysis
- Compares regression models and saves the best pipeline for reuse
Gesture Hill Climb
Original webcam-controlled driving game with real-time hand-gesture input.
- MediaPipe hand landmarks drive acceleration, braking, and jumping
- Calibration and keyboard fallback make the experience testable without a webcam
FitAI Fitness Coach
Fitness and wellness prototype with AI-guided plans, tracking, and coaching flows.
- Workout, nutrition, hydration, goal, and progress-management interfaces
- Backend supports AI assessment, chat, readiness, and recommendation endpoints
From raw data to a usable result.
Let’s build something useful.
Open to AI/ML internships, data-science roles, and applied projects where data can improve a product or decision.