10 pcredits
This Quest merges a short AI/ML code sprint with an advanced Work-Integrated Learning Path (WILP). Participants implement or enhance an AI/ML pipeline targeting a recognized DRR or DRI issue—such as climate-driven vector-borne disease spikes, population displacement predictions, or multi-hazard synergy forecasting. The emphasis is on building an ethically grounded solution with disclaimers around interpretability, data constraints, and local context.
Key Outputs
- AI/ML Pipeline: A containerized or script-based solution addressing a targeted DRR or DRI challenge
- Performance & RRI Metrics: Documenting how well the model handles real or test data, plus disclaimers for underrepresented areas
- WILP Module Completion: Confirming participant’s advanced learning milestone, bridging theory and practice
11 Steps
- WILP Enrollment: Officially join or confirm your advanced AI track, awarding initial eCredits for commitment
- WILP Enrollment: Officially join or confirm your advanced AI track, awarding initial eCredits for commitment
- Data Collation: Gather relevant time-series or geospatial sets from recognized open data archives, ensuring advanced container-based ingestion is allowed
- Model Architecture Sketch: Outline your approach (temporal RNN, ensemble-based random forest, or advanced gradient boosting) and desired outcome metrics
- Initial Training: Use partial data subsets to refine hyperparameters, tracking results in a local or cloud environment
- Interpretability Tools: Apply partial post-hoc checks (like integrated Gradients, LIME) to highlight which features dominate predictions, awarding partial pCredits for advanced ML
- Local Stakeholder Feedback: If feasible, present an alpha model demonstration to relevant community, verifying practical acceptability and disclaimers
- Refinement: Incorporate feedback from both domain experts and WILP mentors, adjusting the pipeline or disclaimers
- Peer Validation: Post your final model code for cross-check in the platform forum, awarding partial validation credits if sufficiently improved
- WILP Reflection: Write a short concluding note describing lessons, model performance achievements, and RRI-based disclaimers
- Integration & Acknowledgment: Merge your code, disclaimers, and WILP reflections into the official DRR/DRI library, awarding final credits once fully approved