DC Project Design Template - RAISE DN
RAISE Scope Reminder
Each DC project must demonstrate that AI development is INTEGRAL to the science.
DC must either develop or significantly participate in developing innovative AI systems, models, tools or methodologies for their scientific domain.
The AI component must substantially innovate how scientific information is analyzed.
NOT acceptable: Pure CS/AI development without domain science; Using existing AI tools instrumentally.
DC Project Summary Table
| Field | Details |
|---|
| DC Number | DC[X] |
| Project Title | |
| Host Institution | |
| Host Country | |
| Supervisor | |
| Co-Supervisor(s) | |
| Scientific Domain | |
| AI Innovation Area | |
| Duration (months) | |
| Planned Start Month | |
| Secondment 1 (Institution, Country, Duration, Purpose) | |
| Secondment 2 (Institution, Country, Duration, Purpose) | |
| Non-academic sector time (months) | |
1. Research Objectives
[Describe 2-3 specific, measurable research objectives]
RAISE Alignment Check:
2. State of the Art
[Current gaps in both the scientific domain AND in AI methods for this domain]
3. Methodology
3.1 AI Innovation Component
[What new AI system/model/tool/methodology will be developed?]
3.2 Domain Science Component
[What scientific questions will be answered?]
3.3 Integration
[How are the AI and domain science components inseparable?]
4. Expected Results
- Scientific publications: [target number and venues]
- AI tools/software: [describe deliverables]
- Datasets: [describe any datasets to be created/curated]
- Domain impact: [specific scientific advancement]
5. Training Plan
5.1 Research Skills
[Domain-specific and AI-specific research training]
5.2 AI in Science Training (MANDATORY for RAISE)
[Dedicated doctoral-level AI-in-science training this DC will receive]
5.3 Transferable Skills
[Communication, entrepreneurship, IP, project management]
6. Supervision Arrangement
| Role | Name | Institution | Expertise |
|---|
| Main Supervisor | | | |
| Co-Supervisor | | | |
| Industrial Mentor | | | |
7. Secondment Plan
| Period | Institution | Country | Duration | Purpose |
|---|
| Secondment 1 | | | months | |
| Secondment 2 | | | months | |
8. Risk Assessment
| Risk | Likelihood | Impact | Mitigation |
|---|
| L/M/H | L/M/H | |
| L/M/H | L/M/H | |
9. Mobility Rule Compliance
Suggested DC Project Clusters for AI-Driven Scientific Discovery
Cluster 1: AI for Physical Sciences (DC1-DC4)
- DC1: [Placeholder: AI-driven materials discovery] — Develop generative AI models that propose and screen novel functional materials by jointly learning from crystal structure databases and quantum-mechanical property calculations, accelerating the design-synthesis loop beyond exhaustive DFT screening.
- DC2: [Placeholder: Physics-informed neural networks for simulation] — Construct physics-informed neural networks that embed conservation laws and symmetry constraints as hard architectural priors, enabling surrogate simulations of complex fluid or plasma dynamics at a fraction of classical solver cost.
- DC3: [Placeholder: AI for astronomical data analysis] — Design self-supervised deep learning pipelines that detect and classify rare transient events (e.g., gravitational microlensing, fast radio bursts) in real-time streams from next-generation sky surveys such as LSST/Rubin.
- DC4: [Placeholder: Quantum-AI hybrid methods] — Develop hybrid quantum-classical machine learning algorithms that leverage near-term quantum processors to enhance the expressivity of AI models for quantum chemistry ground-state and excited-state prediction.
Cluster 2: AI for Life Sciences (DC5-DC8)
- DC5: [Placeholder: AI for drug discovery and molecular design] — Create multi-objective generative AI architectures that simultaneously optimize drug-likeness, target affinity, and ADMET properties, with active-learning loops guided by automated wet-lab assays to close the in silico–in vitro gap.
- DC6: [Placeholder: AI-powered genomics and precision medicine] — Develop transformer-based foundation models pre-trained on large-scale multi-omic cohorts to disentangle genetic, epigenetic, and environmental contributors to complex disease, producing patient-stratification tools for clinical translation.
- DC7: [Placeholder: Computer vision for microscopy/pathology] — Build weakly supervised and self-supervised vision models that extract quantitative morphological biomarkers from whole-slide histopathology images, linking spatial tissue architecture to molecular phenotypes and patient outcomes.
- DC8: [Placeholder: AI for protein structure prediction and design] — Extend structure-prediction AI frameworks (beyond static snapshots) to model conformational ensembles and allosteric dynamics, enabling the de novo design of proteins with programmable functional transitions.
Cluster 3: AI for Earth & Environmental Sciences (DC9-DC11)
- DC9: [Placeholder: AI for climate modeling and prediction] — Develop neural emulators trained on high-resolution Earth-system model ensembles to deliver kilometer-scale regional climate projections at global-model cost, with uncertainty quantification methods calibrated against observational reanalysis data.
- DC10: [Placeholder: AI-driven environmental monitoring] — Construct multimodal AI systems that fuse satellite imagery, IoT sensor streams, and citizen-science observations to produce near-real-time maps of biodiversity indicators and ecosystem health at landscape scale.
- DC11: [Placeholder: AI for sustainable resource management] — Develop reinforcement-learning and causal-AI frameworks that optimize spatiotemporal resource-use decisions (water, energy, land) under climate uncertainty, integrating domain knowledge from hydrology and agronomy as inductive biases.
Cluster 4: AI for Social Sciences & Cross-cutting (DC12-DC15)
- DC12: [Placeholder: AI for social data analysis and NLP] — Design domain-adaptive large language models fine-tuned on longitudinal social-science corpora to extract latent behavioral patterns, opinion dynamics, and cultural shifts from heterogeneous text and network data with rigorous uncertainty estimates.
- DC13: [Placeholder: Explainable AI for scientific decision-making] — Develop post-hoc and ante-hoc explainability methods tailored to scientific AI models (e.g., graph neural networks, neural ODEs), producing human-interpretable explanations that align with disciplinary reasoning conventions and support hypothesis generation.
- DC14: [Placeholder: AI ethics and responsible innovation in science] — Produce a computational ethics framework and associated auditing toolkit for evaluating bias, fairness, reproducibility, and societal impact of AI systems deployed in high-stakes scientific domains, grounded in empirical case studies across RAISE partner disciplines.
- DC15: [Placeholder: AI-driven automated scientific workflows] — Architect and validate an AI-orchestrated autonomous experimentation platform that closes the hypothesis-experiment-analysis loop, combining large language model reasoning for experimental planning with robotic lab integration and Bayesian optimization for adaptive design-of-experiments.