Accueil / Experts / BACHAR Hamza
Deep Learning and MLOps expert
Deep Learning Machine Learning MLOps Computer Vision NLP LLM GenAI
.
contribute towards achieving the team's objectives.
WORK EXPERIENCE
Technip
Oct 2022 - Present Paris
Novartis
Dec 2021 - Sep 2022 Basel
Hermes
May 2020 - Nov 2021 Paris
ENGIE
Jun 2020 - Apr 2021 Paris
OuiSNCF
Feb 2019 - Feb 2020 Paris
Thales
Apr 2017 - Feb 2019 Sophia Antipolis
SR DEEP LEARNING / AI ARCHITECT
Solution: Databricks, AzureML, MLFlow, DataIku, ADO Pipelines CI/CD/CT/CM, ACR, AKS, MLOps, LLM, RAG, Generative AI, Vector Search. Description: Building From scratch an MLOps framework for AI/ML teams. to standardize the ML Life cycle and accelerate the product to GO Live.
• The current state inventory of all Machine Learning projects
• Designed physical and global architectures for efficient AI/ML operations
• Automated ML life cycle through Azure DevOps pipelines and Databricks webhooks for seamless deployment and testing
• Build Data Drifting detection and Model retaining pipelines.
• Leading and assisting team members with technical and functional tasks.
• Build a chatbot search engine using Azure Open AI GPT3.5-Turbo with RAG
MLOPS WITH DATABRICKS AND AZURE PIPELINES
Solution: Databricks, AzureML, MLOps, MLFlow, ADO Pipelines CI/CD/CT/CM, ACR, AKS Description: Building MLOps Framework for AI ML teams and automation production rollout process
• Building the physical and logical architectures to include Databricks in the current solution.
• Automating pipeline with Azure DevOps, by building docker image (ACR) testing and deploying the model in an AKS cluster.
• Managed ML lifecycle & triggered CI/CD/CT/CM pipelines for automatic model transitions using webhooks.
SR DEEP LEARNING ENGINEER / VISUAL SEARCH
Solution: SageMaker, TensorFLow2.0, Pytorch, Pandas, snowflake, and Sklearn
Description: Optimized item assignment process in the after-sales department using AI.
• As the tech lead, planning and defining the technical solution in an AWS environment.
• Conducted benchmarking and research to develop innovative Deep Learning solutions for Computer Vision
• Design the architectural solution applicative and data flow
• Building technical backlog tasks and leading juniors in the team
• Presenting, and discussing the solution and the results with non-technical final users
• Functional and technical Documentation of the solution.
TECH LEAD DATA SCIENTIST
Solutions: Databricks, AzureML, MLFlow, MLOps, ADO, Pytorch
Description: Design, develop, and deploy a fully automated NLP project. For different departments, legal and marketing.
• Client-facing, proposing technical solutions for pain points in different departments
• Collecting data from multiple sources, internal/external.
• Finetuning LLM for a legal-powered AI for French documents
• Preparing data, cleansing text, and transforming for
AI model fine-tuning
• Training a Transformer classifier AI model based on
HuML-specified annotations.
• Building Continuous Integration, Deployment, and
Training Pipelines.
CONSULTANT DATA SCIENTIST
Solutions: Tensorflow, Keras, pySpark, SparkML, Sklearn, xgboost
Description:
• Developed innovative tools for ouiSNCF, including PIC2TRIP and CrowdPred
• Coached junior data scientist and contributed to project qualification/design
• Collected and scraped data for analysis
• Pre-processed, cleaned, and filtered data for accurate modeling
• Model Training and evaluation and experiments tracking.
• Serve model using flask.
DEEP LEARNING ENGINEER
Consultant Data Scientist
Solutions: TensorFlow, Keras, Pandas, OpenCV, Scikit-learn
Description: Cell-final inspection guides the visual inspection of photovoltaic cells by artificial intelligence (Neuron Networks).
• Balance and increase the learning database by smoothly generating images of defective cells.
• Pre-process cell images for optimal analysis.
• Trained Deep Learning model to accurately score cell defect probability
• Deploy the model with a continuous improvement strategy.
• Implemented continuous improvement strategy for model training&deploymentntt