ICLR 2024Featured
Tokenization Strategies for Morphologically Rich Semitic Languages
A. Kebede, S. Johnson, M. Haile.
PDFCode
Our ResearchAdvancing AI for
Advancing AI for
Real-World Impact
From language technology to climate modeling, explore the research powering equitable AI solutions across Africa and beyond.
Focus Areas
Core Research Domains
We tackle Africa’s most pressing challenges through three interconnected research pillars.
Natural Language Processing
Language AI (NLP)
Developing Large Language Models (LLMs) and translation systems for low-resource African languages, preserving cultural heritage and enabling digital access.
LLM Fine-tuningMachine TranslationMorphological AnalysisSpeech Recognition
Policy & Ethics
Governance & Legal AI
Streamlining judicial processes, digitizing legal archives, and creating transparency tools for governance through intelligent document processing.
Document IntelligenceLegal NERPolicy AnalysisTransparency Tools
Food Security & Sustainability
Agriculture & Climate AI
Predictive modeling for crop yields, pest detection, and climate resilience strategies to support food security across the continent.
Crop Disease DetectionYield PredictionClimate ModelingSatellite Imagery
Our Work
Featured Projects
Highlights from our current initiatives driving AI impact.
NLPActive
Project Fidel: Amharic LLM
A foundational Large Language Model specifically fine-tuned for Amharic and Ge’ez script languages. This project aims to democratize access to generative AI tools for over 50 million speakers in the horn of Africa.
Read Case Study
AgricultureActive
CropGuard AI
Using computer vision on satellite and drone imagery to detect early signs of rust disease in wheat crops. Early warning systems deployed in three regions have saved an estimated 15% of annual yield.
Read Case Study
HealthcareIn Partnership
Predictive Diagnostics Engine
Novel deep learning architectures to analyze medical imaging data. In partnership with major research hospitals, the collaboration resulted in a 14% improvement in early detection rates for rare pathologies.
Read Case Study
ClimateIn Partnership
Climate Modeling at Exascale
Utilizing high-performance computing clusters to refine climate prediction models. By integrating reinforcement learning, we reduced computational costs by 40% while maintaining high-fidelity simulation accuracy.
Read Case Study
Publications
Published Research
Peer-reviewed papers from our lab, advancing the state of the art.
AllNLPComputer VisionAgricultureHealthcare
NeurIPS 2023
Few-Shot Learning for African Crop Disease Detection
B. Tadesse, R. Smith.
PDFDataset
ACL 2023
Cross-lingual Transfer for Low-Resource Ethiopian Languages
M. Haile, A. Kebede, L. Tesfaye.
PDFCodeDemo
CVPR 2023
Satellite-Based Crop Monitoring with Self-Supervised Vision Transformers
B. Tadesse, K. Ayele.
PDF
Our Research Methodology
From hypothesis to deployment, our iterative cycle ensures excellence at every stage.
1
Identify
Partner with industry to pinpoint high-impact inefficiencies.
2
Research
Develop novel algorithms and test against state-of-the-art benchmarks.
3
Prototype
Build minimum viable models (MVM) for controlled environment testing.
4
Deploy
Scale solutions with continuous monitoring and reinforcement learning.
Collaborate With Us
Interested in our research or looking to partner on a project? We welcome collaborations with academic institutions, industry partners, and government organizations.