Infor

Intern – Automatic Hyperparameter Tuning for Demand Forecasting

4/5-month duration from February 2024

Mission/Challenges:

The intern will work on implementing a mechanism for automatic hyperparameter tuning in the AI Engine, which is responsible for ML-based demand forecasting. The goal is to make the engine autonomous and adaptive to each customer's needs by automatically setting hyperparameters. This will involve creating a system that uses techniques like holdout mode to calculate error metrics, avoid overfitting and underfitting problems, and adapt to new projects. Additionally, the intern will need to devise a strategy for the engine to monitor errors during weekly batches and adjust parameters, if necessary, limited to a set number of runs each week.

Skills and Requirements:

• Strong programming skills, preferably in Python.
• Knowledge of machine learning techniques, especially factorization machines.
• Understanding of hyperparameter tuning methods, such as grid search or Bayesian optimization.
• Familiarity with holdout validation techniques and error metric calculation.
• Ability to create scripts for automated parameter adjustments. Analytical thinking and problem-solving skills