Data • AI Engineer
Automated classification of aviation incident reports, reducing manual triage time by ~90% (from ~30 min to <5 sec per report)
End-to-end LLM architecture for BEA accident report analysis: classification, HFACS extraction, semantic RAG and weak signal detection (all-in-one).
End-to-end RAG pipeline for intelligent data extraction and hallucination-free LLM querying
Architected a complete system including PDF chunking, high-dimensional embedding generation, FAISS vector storage, and LLM querying with contextual constraints.
Automated KPI extraction from financial data, reducing manual research time by 30%
Financial chatbot automating the extraction of critical KPIs. Pandas pipeline structuring 3 years of financial data into interactive metrics.
Classification model for glaucoma detection with 92% F1-Score and expert validation
5 feature extraction approaches and 4 classification models for automated glaucoma detection. Validated by an expert ophthalmologist.
Deep neural network for image classification with an optimised data pipeline
Complete data pipeline with optimised image preprocessing. Design and training of a deep neural network for classification.
Vectorised implementation without predictive ML libraries, built from scratch using OOP
OOP learning model without external libraries. Vectorised Gradient Descent implementation via NumPy. Training and testing on real-world data.
French
Native
English
C1
Chinese
B1
Arabic
B2
I am open to internship and work-study opportunities and interesting projects in AI and Data Science.