Foundation Model Evaluator 

Problem statement

Stellantis, renowned for its innovative spirit, is pushing into the practical application of Generative AI. At the heart of this technological frontier is the challenge of selecting the best-fitting foundation model that aligns with specific use cases – taking into account not just performance but also other factors like cost-effectiveness and reliability.  

Join us in refining the process of choosing the right foundation models for practical use cases, which are pivotal in the ever-evolving landscape of AI technologies. 

Your task

To enhance the selection process for available foundation models Stellantis identified two main areas of improvement. You can tackle either one or combine both to take part: 

1. Performance Assessment & Selection: Propose a solution or feature that aids in selecting the most suitable foundation model for various practical use cases, enhancing the efficiency of the selection process and the quality of the results for each use case. 
2. Performance Evaluation Post-Deployment: Develop a robust method or feature to evaluate the chosen models’ performance, specifically addressing internal challenges and e.g. incorporating user feedback to monitor and improve the models continuously. 

Your task is to develop at least a detailed concept or if possible, a first prototype to apply. The goal is to build a functional POC in the next 6-month after the initial deadline. The approaches you propose will form the cornerstone of Stellantis's broader strategy to enhance internal operations through GenAI, giving you a unique opportunity to push your career or venture.   

For more details about the Key Guidelines & Submission Criteria click on the tab "Submission".