I'MDr Esso-Ridah BLEZA

I'm an innovative, developper and designer customer whith:
Strong experience in methodological developments with recent techniques in quantitative analysis, advanced statistical modeling, natural language processing (NLP) models, time series forecasting, Machine learning and deep learning models to effectively exploit strategic and business ideas oriented to the solutions of real application problems

Technology Using:
Python (advanced), R, SQL, dynamoDB, Spark, Pyspark , AWS, SageMaker, Jupyter Notebooks, Numpy, Pandas, Scikit-learn, caret, Keras, Tensorflow Matplotlib, Seaborn, etc.

Strong understanding of machine learning algorithms, conceptualizing and building efficient data processing pipelines by training, testing and deploying high quality models in production capable of running at scale

Extensive experience in Data Wrangling and Feature Engineering techniques

I was responsible for the MLOps Enginering for data science role during my last experience:
Architected and implemented end to end solution for model deployment, from data ingestion to model deployment and inferences.The use case was the localized recognition of pollens in the air with the allergy risk levels incurred by the Live Pollen application since March 2022 in France. This was done on full AWS cloud computing.

AWS Services and tools used:
Lambdas, Stepfunctions, S3,Docker, CloudFormation, EC2, and CloudWatch, SageMaker, DynamoDB, Postgres, etc.

Onboarded the team’s first use case for an MLOps platform which is developed to take machine learning models from PoC to production. I will developed the solution individually or will be a mentor as AWS consultants for specifically:
• Built Docker container with the required environment and pushed it to AWS ECR.
• Created AWS Sagemaker jobs required for model training, tuning and evaluation.
• Automated the model training pipeline/process using AWS step functions.
• Create and manage Nosql data base using S3, DynamoDB.

I'am interested in high stream data compressing of data by proposing methods for innovate companies. These data constitute important spatio-temporal flows characterized by variables that can be redundant, and whose distributions are asymmetric (non-Gaussian) and with zero inflation.

Research contribution:
segmentation method of a data stream using a dynamic programming algorithm based on a piecewise linear approxima- tion and a complexity criterion (minimum description length)
unsupervised classification based on kernel methods with distance optimization for zero-inflated data with heavy tail.
original supervised pollen detection and classification algorithms.

These methods have been implemented in applications like the (non-intrusive) monitoring of old people for a better management in case of loss of autonomy or accident (fall, illness, lack of activity, etc) or study concerns the prevention of allergy risks linked to the diffusion of pollens in the outdoor air. The algorithms developed have been deployed and produce and broadcast preventive alerts on mobile applications

I offer support to improve skills. Any form of support you wish is possible whatever your level, from high school to Master's degree. Explanation of statitics theoretical bases and ML fundamental concepts, correction of python and R codes for practice, contribution to the improvement of algorithm scores, etc.

Spatiality area:
AI & ML algorithms, data analysis, python an R for data science, statistics, probability, mathematics, statistics.

I am assigned as an assistant during the practical work of the Master's students' academic projects in machine learning. I was Trainer and mentor on machine learning projects for students of the Master in Statistics of the University of Rennes 1.