About Me

In a nutshell, I represent a vector in a three-dimensional space, defined by the coordinates of entrepreneurship, AI R&D for real-world impact, and amateur astrophotography as a hobby.

I am focusing on generating valuable insights for business stakeholders through innovative AI models, with a particular emphasis on massive data collection, web scraping, NLP, knowledge graphs,graph learning, LLMs, and machine learning.

I graduated from the Ecole des Mines de Saint Etienne (EMSE), securing not only my engineering degree but also attaining a Master's in Advanced Mathematical Imaging and Pattern Detection.

I have worked as :
  • R&D Engineer in AI @ AUVALIE INNOVATION (november 2022 - now)
  • Hyperspectral Image Recognition Engineer @ CNES (april 2022 - october 2022)
  • Natural Language Processing Engineer @ AUVALIE INNOVATION (june 2021 - february 2022)
  • Founder @ COSMIC SYMPHONY (september 2021 - present)

    • Skills

    Programming Languages: Python, SQL, Cypher.

    Programming Architecture Patterns and Concepts: Hexagonal Pattern, SOLID Principles, OOP (Object-Oriented Programming), Domain-Driven Design, Microservices.

    Databases: PostgreSQL, Neo4J, MongoDB.

    MLOps: Docker, Kubernetes, Microservices.

    Data Visualization: Matplotlib, NetworkX, D3.js.

    Code Repository: GitHub.

    IDE: Visual Studio Code, Jupyter Notebook.

    AI: Natural Language Processing(NLP), Large Language Models(LLMs), Graph Learning, Graph Embeddings, Word Embeddings, Massive Data Analysis, Information Retrieval, Business Insights Prediction, Machine Learning.

    Projects I've Built

    These are some of my featured projects that I have built.

    iText2KG

    iText2KG is an open-source Python toolkit that leverages large language models to convert unstructured text into a continuously evolving knowledge graph in a zero-shot, topic-agnostic manner. By eliminating the need for domain-specific training data or manual ontology design, iText2KG enables organizations to unlock insights from the 80% of data trapped in text, whether scientific literature, web content, or resumes, while its incremental update and hallucination-mitigation strategies ensure the graph stays accurate and up to date, making it a powerful plug-and-play solution for knowledge-driven analytics and AI applications.

    Python, LLMs, Knowledge Graphs, NLP, Neo4j, LangChain.

    Manazel

    Manazel is an AI-driven Python package that uses 13 years of naked-eye lunar crescent observation data from Morocco to refine the widely used ODEH criterion and applies a logistic regression model to predict crescent visibility with 98.83% accuracy, thereby automating and standardizing the traditional, highly variable process of Hijri-month determination (critical for scheduling religious, cultural, and administrative events) and reducing uncertainty in calendar decisions.

    Python, Machine Learning, Logistic Regression, Astronomy, Data Analysis.

    Jovian Image Processor

    Jovian Image Processing, is an image processing web application coded with Python for Back-End and Angular for Front-End, to edit or process NASA JunoCam raw images in order to enhance storms shape and different zones.

    Python, OpenCV, Scraping, Adaptative Histogram Equalisation (AHE), median filter, noise reduction, sharpen filter

    HyperCheck

    HyperCheck is an AI application that helps space engineers to reduce satellite failures through automatic detection of organic materials which have the disadvantage of out-gassing under vacuum and contaminating the satellite detectors.

    Python, OpenCV, Hyperspectral Image Processing, Independent Components Analysis, Principal Components Analysis, sklearn, skimage, KMeans, Automatic Segmentation, supervised classification, 2DCNN, Fast3DCNNs , Keras, Parallel computing, GUI, PyQt5, QtDesigner.

    Mind Mapping Application

    This is a NLP based application which creates automatically mind maps for technical documents very rapidly and efficiently.

    Python, Natural Language Processing, Topic Modeling, Graph theory, NTLK, Spacy, BERT, VueJS, fastAPI, axios, docker, Github.

    PyStarTrails

    PyStarTrails is a free python package which could be used by astrophotographers and photographers to create stunning images of the night sky by generating star trails from a sequence of images.

    Python, Image Processing, OpenCV, Image Blending, Github.

    Publications

    Lairgi, Y., et al. "itext2kg: Incremental knowledge graphs construction using large language models." International Conference on Web Information Systems Engineering. Singapore: Springer Nature Singapore, 2024. READ

    Hajji, A., Lairgi, Y., Lachhab, A., & Abbou, A. (2020). Impact of waste recovery on the COVID-19 propagation. Mediterranean Journal of Chemistry, 10(8), 744-751. READ

    Lairgi, L., Lagtayi, R., Lairgi, Y., Daya, A., Elotmani, R., Khouya, A., & Touzani, M. (2022). Forecasting Heating and Cooling Energy Consumption by Seasonal ARIMA Models. In International Conference on Digital Technologies and Applications (pp. 65-74). Springer, Cham. READ

    Bourcier, F., Walter, P., Faye, D., Rioland, G., Infante, F., Gouisset, E., ... & Lairgi, Y. (2022, October). Performance review of a UV/vis/IR fluorescence hyperspectral camera to detect contamination on spacecraft during integration. In Space Systems Contamination: Prediction, Control, and Performance 2022 (Vol. 12224, pp. 336-346). SPIE. READ

    Achievements

    Get in touch ?

    Feel free to contact me, whether you have a question or just want to discuss about my previous projects. I am always open to any job updates :)