Skip to content

Data Analyst Intern

  • Hybrid
    • Utrecht, Utrecht, Netherlands
  • Product & Technology

Job description

Internship Proposal

Background

Food waste is a major global issue, with significant environmental and economic consequences. Many restaurants have already implemented various actions to minimize food waste, such as portion control, dynamic pricing, and improved inventory management. However, identifying the most effective actions and tailoring them to specific contexts remains a challenge.

At Orbisk, we aim to enhance our food waste reduction strategies by leveraging advanced technologies. The goal of this internship is to research and develop a Large Language Model (LLM) that can suggest actionable strategies based on the food waste data of our clients. The intern will explore data-driven approaches to recommend optimal actions, with the potential for the model to become self-learning. This means the model will propose actions, the user will select the preferred one, and this feedback will be used to improve future suggestions.

Assignment

The intern will focus on the following tasks:

  1. Data Collection: Gather data on potential food waste reduction actions from various sources, including but not limited to:

    • Existing Large Language Models

    • Orbisk’s existing action centers

    • Input from our food waste coaches

    • Resources from the Zero Hunger Lab

    • Online resources

  2. Model Development: Research and develop an LLM capable of analyzing client-specific food waste data and suggesting tailored actions.

  3. Self-Learning Mechanism: Design a feedback loop where the user’s selection of the best action informs the model, enabling it to refine its suggestions over time.

  4. Evaluation and Testing: Assess the model’s performance by comparing suggested actions against actual outcomes, refining the model based on these results.

  5. Visualization and Reporting: Present the findings in a clear, actionable format, aiding stakeholders in making informed, data-driven decisions.

Expected Outcomes

  • A functional LLM that suggests food waste reduction actions based on client data.

  • A self-learning mechanism that improves action recommendations through user feedback.

  • A comprehensive dataset enriched with potential actions and corresponding outcomes.

  • A report and visualizations showcasing insights, model performance, and recommendations for optimizing food waste management.

  • Potential contributions to Orbisk’s broader efforts in optimizing food waste management strategies

or