Engineering Practice Problem
Develop ML-powered prediction system estimating Supercharger wait times using occupancy data, temporal patterns, and external factors. Build REST API suggesting alternative stations when waits exceed 20 minutes.
Problem Statement
Submission Guidelines
Create a prediction system and API that estimates Supercharger wait times for the next 4 hours in 30-minute intervals. Build a model that incorporates temporal features (hour/day/month), location features (highway vs urban, proximity to attractions), and external factors (weather, holidays, local events). Since real Tesla data isn't available, create a synthetic dataset generation system that simulates realistic usage patterns based on publicly known information about Supercharger locations and EV charging behavior. Design a REST API that accepts location and estimated arrival time, returning predicted wait time with confidence bounds and suggesting alternatives if wait exceeds 20 minutes.
Deliverables
Submit ALL of the following:
Machine learning or statistical model for wait time prediction with trained model file
Synthetic dataset generation code simulating realistic Supercharger usage patterns
REST API implementation with full documentation (FastAPI, Flask, Express.js, or similar)
Jupyter notebook showing model training, feature engineering, and evaluation with visualizations
API testing suite with at least 10 example requests/responses
Performance dashboard or report showing prediction accuracy metrics across different scenarios
System architecture diagram (1 page) showing data flow and components
This practice problem is suitable for:
Judging Criteria
Prediction accuracy (30%) – Model performance metrics (MAE, RMSE) on test scenarios and edge cases
Feature engineering (20%) – Thoughtfulness and effectiveness of predictive signals and data representation
API design (20%) – Usability, documentation quality, error handling, and response format
System design (15%) – Scalability considerations and architecture for real-world deployment
Data generation (15%) – Realism of synthetic dataset and coverage of diverse charging scenarios
Want expert feedback on your submission?
Get detailed, personalized feedback from experienced judges on your submission—what worked, what didn’t, and how you can improve. Here is what's included:
Suggestions to make your submission/idea better.
Written comments on your strengths and areas of improvement.
Written comments on your strengths and areas of improvement.
A detailed assessment of your submission with scores for different judging criterias.
₹999
