Ndèye Maguette Mbaye

- Certifications: Post-training of LLMs; DeepLearning.ai, Langchain and Langgraph Specialization – Coursera

– Generative AI Project: Seetlu: RAG Agent with latency ≤3s and 90%+ accuracy (LLM-as-judge evaluation), which answers questions by retrieving and summarizing relevant information from my professional documents.

– Develop 5+ ML/DL models (Transformers and LLMs) applied to multimodal data of 15K+ examples for prediction tasks with 0.78 AUC (+10% compared to benchmarks), Conduct various embedding strategies (Time embeddings with time2vec, embeddings for continuous variables, contextual embeddings), which improved performance by +8%,

– Implement reproducible and scalable AI pipelines, integrating data preparation, training, evaluation, performance monitoring, and interpretability,

– Collaborate with engineers, PhDs, and end-users to validate results,

– Write technical and educational deliverables for non-technical teams

– Develop Transformer-based models for electronic health records using different types of integration,

– Compare and evaluate embedding strategies to improve predictive performance (+7%).