DeepFrog Applied Lab
Where applied AI research meets real-world deployment.
Our Labs team turns frontier research into production-grade systems. We publish openly, collaborate with universities, and maintain open-source tooling used by thousands of engineers around the world.
Research Focus
Six research programs. One engineering backbone.
Foundation Models
6 papersEfficient fine-tuning, instruction alignment, and domain specialization of open-weight LLMs.
Multimodal RAG
4 papersRetrieval-augmented generation across text, image, and structured data for regulated domains.
Medical Imaging
9 papersPrivacy-preserving vision models for radiology, pathology, and federated learning at NHS scale.
Responsible AI
5 papersBias auditing, interpretability, and policy-aware evaluation aligned to the EU AI Act.
Predictive Systems
4 papersTime-series foundation models and anomaly detection in financial and industrial data.
Edge & On-Device AI
4 papersQuantization, distillation, and deployment of compact models to factory and automotive hardware.
Recent Publications
Selected papers from the Lab.
- Read paper →
NeurIPS 2025
Private RAG: Differentially-Private Retrieval for Enterprise LLMs
Varga E., Ramanathan P., et al.
- Read paper →
MICCAI 2025
Federated Chest X-ray Triage across 40 Hospitals
Ramanathan P., Al-Hassan Z., et al.
- Read paper →
ICML 2025
Compression-Aware Fine-Tuning of 70B-scale Foundation Models
Brandt L., Varga E., et al.
- Read paper →
CVPR 2025
Self-supervised Defect Detection for Semiconductor Manufacturing
Tanaka H., et al.
- Read paper →
FAccT 2025
Auditable Model Cards for the EU AI Act
Al-Hassan Z., Okoye M., et al.