About

Professional experience

• Senior Data Scientist, Data Science Team, AXA France

  June 2021 - Present

  • Developed an LLM-based RAG system, reducing help-desk workload by providing instant answers to insurance agents.

  • Led workshops with ML Engineers, Data Engineers, and Data Scientists to develop the MLOps Best Practices Guide for AXA.

  • Created a Cookiecutter template for new ML projects to accelerate time-to-market (TTM) at AXA France.

  • Developed text extraction AI systems using computer vision, processing over 6M documents in batches with PySpark, reducing processing time by 30%.

  • Created reusable Python packages for document processing, enabling rapid experimentation with various algorithms.

  • Automated the indexing of 70% of incoming documents through a retrainable document classification AI system, improving efficiency.

  • Built reusable Azure ML pipelines for continuous training, ensuring models are always up-to-date in the Model Registry.

  • Designed CI/CD/CT pipeline templates in Azure DevOps, improving project delivery timelines.

  • Led the hiring process and onboarding of new team members, facilitating efficient knowledge transfer and team integration.

  • Coordinated the annotation process with subject matter experts to ensure high-quality datasets across various projects.

  • Led AXA's 'Python COP' initiative, organizing monthly talks on topics such as testing, web scraping, and code quality.

  • Conducted training sessions on Python and data manipulation for over 30 AXA collaborators, enhancing their skills.

• Graduate Data Scientist, Data, Fraud, Waste & Abuse Team, AXA France

  May 2020 - June 2021

  • Engineered an AI-driven Fraud, Waste, and Abuse Detection Tool, achieving savings exceeding 150k euros in 2022.

  • Designed and trained multiple fraud detection models, significantly improving detection accuracy.

• Data Scientist, Shift Technology

  April 2019 - March 2020

  • Applied NLP techniques to cluster claim managers' notes, revealing emerging fraud patterns through advanced visualization.

  • Developed and deployed new AI models for Shift's fraud detection solution, enhancing client offerings.

• Research Internship, U2IS - ENSTA Paris

  May 2018 - August 2018

  • Created a synthetic image dataset by simulating an agent in an in-house environment, utilizing ray tracing for realistic image rendering.

  • Researched catastrophic forgetting in incremental learning of principal components of generated images.

Projects

• mlflower: Lightweight orchestration for mlflow projects

  November 2023 - Present

  An orchestration tool that enhances MLflow projects by providing built-in multi-step workflow management.

  • Utilized Directed Acyclic Graphs (DAGs) for managing complex ML workflows, ensuring logical task execution.

  • Implemented topological sorting to optimize task order based on dependencies, enhancing workflow efficiency.

  • Enhanced MLflow projects with clear dependency management for multi-step experiments.

  • Integrated visualization tools for intuitive understanding of workflows and dependencies.

• pysira: Publish resumes in different formats

  February 2023 - Present

  A CLI tool for exporting 'jsonresume' files to various formats (HTML, TeX, PDF) and languages.

  • Developed a CI/CD pipeline using GitHub Actions to automate resume publishing in multiple formats and themes.

  • Enabled export of 'jsonresume' files to HTML, TeX, PDF, and other formats.

  • Demonstrated functionality with a live resume: hmiladhia.github.io, hmiladhia.github.io/cv.pdf.

• splitter: Document Processing Tool

  December 2021 - Present

  An open-source package that provides tools for processing various document types (PDF, TIFF, etc.) into images and extracting text using Python.

  • Supports multiple document formats for versatile processing.

  • Customizable architecture with plugin support for extended functionality.

• nbmanips: Split, merge and convert IPython Notebooks

  May 2021 - Present

  An open-source toolset for manipulating Jupyter Notebooks via Python or CLI.

  • Created a Python package and CLI for splitting, merging, and converting Jupyter Notebooks.

  • Set up CI/CD pipelines using GitHub Actions for linting, testing, and publishing to PyPI.

  • Contributed to the open-source community by providing tools for Jupyter Notebook management.

• Spot-Language: Programming Language Detection

  January 2020 - Present

  A classification model for detecting programming languages in code snippets.

  • Assembled a training dataset from public GitHub repositories on Github

  • Built an ML experiment to train a classification model for programming language detection.

  • Enhanced model interpretability using LIME for better insights into predictions.

  • Deployed a demo web app using Flask to showcase the ML model.

• Dmail: Markdown e-mails

  April 2020 - May 2020

  A Python package for composing and sending emails in markdown format.

  • Developed a Python package for composing and sending emails in markdown format.

  • Established CI/CD pipelines with GitHub Actions for linting, testing, and publishing to PyPI.

  • Achieved over 100k downloads, contributing to the open-source community.

Skills

Education

Certificates