Riadh BALTI

-Established an automated system to patch missing data in databases, employing Kafka streaming and Kafka Connect for a swift data transfer, resulting in a 20% reduction in data processing time. - Enhanced Spark-based ETL jobs, upgrading migration scripts from Python 2 to Python 3, leading to a 15% improvement in processing efficiency. - Validated data quality using Great Expectations, generating reports and uploading them to an AWS S3 bucket, ensuring a 98% accuracy rate across 100 million records. - Controlled ETL migration across PostgreSQL and Cassandra databases using Spark, Kafka, and Airflow, resulting in a 25% reduction in data transfer latency. - Developed a Monitoring Dashboard for comprehensive data oversight using Apache Superset, increasing monitoring efficiency by 30%. - Design and implementation of PostgreSQL database for efficient storage - Implemented CI/CD pipelines to generate Docker images, subsequently utilized by Airflow 2, and tested fast API endpoints, resulting in a 20% reduction in deployment time. -Web scraping using Node.js, Python requests, and internal tools. Technologies: Python, Node.js, Scala, Kafka, Pandas, Docker, Spark, Cassandra, PostgreSQL, Great Expectations, Gitlab CI/CD, AWS (S3, Redshift, Lambda, RDS)

- Combine raw information for water network from differents sources - Build data systems and pipelines (ETL) to transform raw data (csv, json files). - Prepare database for modeling water supply network through GIS tools - Daily drinking water demand forecasting, artificial neural network. - water leak detection using machine learning algorithms Tools: Python, Apache Airflow , FME, Keras, Numpy, Pandas, Matplotlib,

Multimodal Matching for Fashion products using deep learning methods: consists of using MUTAN, a method for multimodal fusion between Visual and Text using a binilear operation. Consistencies: - Extract the visual attributes using the Resnet-152 network with an attention mechanism - Extract the textual attributes using GRUs similar to RNN using a pre-trained Skip-Thoughts model - Training and validation of the multimodal model, - Classification using the KNN method to give us the products close to a desired product.

• Completion of both Advanced planning and Front End Engineering Design (FEED) of the installation plan of a base camp in Kaga Bandoro and Ndele in Central African Republic (software used Autodesk Civil3D) o Design and calibration of laterite roads o Implementation of stormwater drainage structures (sizing of earthen ditches and canals masonry) o Designing a supply and drinking water system (Sizing pipes and regulating tank) o Environmental impact assessment • Master Plan for the Improvement of Drinking Water Supply in the Rural Areas of Sidi Bouzid (Tunisia)-Phase2 o Classifying variants of water resources o Defining layouts and sizing the necessary infrastructures, (tanks, pumping stations) o Hydraulic single period and extended period simulation using MIKE URBAN • Royal Palace Wetlands Rehabilitation project in AL YAMAMA Castle-Mecca-Saudi Arabia; Software used (Autodesk Civil 3D) o Drawing network layouts o Calibration pipes/ Pipeline length profiles o Coordination plan between different wet networks • Hydraulic study of the extramural rainwater drainage system in a hotel (Cigale-Gammarth-Resort) project area o Evaluation of hydraulic calculation notes of rainwater o Hydraulic modeling using Autodesk Storm and Sanitary Analysis o Plans folder development on Autocad • Royal Palace Wetlands Rehabilitation Project in Mina-Mecca-Saudi Arabia; Software used (Autodesk Civil 3D) o Drawing network layouts o Calibration pipes/ Pipeline length profiles o Coordination plan between different wet networks • Master Plan for the Improvement of Drinking Water Supply in the countryside of Sidi Bouzid (Tunisia)-Phase1 o Mapping the different components of drinking water network in the rural area of Sidi Bouzid using Arcgis o Adopting Cspro to capture and analysis surveys data carried out with inhabitans

Energitic and agronomic valorization Assessement of raw biomasses and impregnated with olive mill wastewater × This research work deals with the energetic and agronomic valorization of two lignocellulosic materials (almond shell and olive solid wastes). • Target: Energy valorization: was carried out through a slow and controlled pyrolysis experiment in an anaerobic environment with the presence of inert gas (nitrogen) to give rise to by-products: Biochars, Bio-oil and Biogas can be exploited as an alternative source of energy (CH4, CO, CO2, and H2) Agronomic valorization: aims on one hand to use the solid residuals "Biochars" as a soil amendment, on the other hand providing crops with additional nutrients beneficial for their growth, in particular potassium. • Results: o Pyrolysis tests have shown that the two raw biomasses presented respectable yields production of biochars and bio-oils. o The biomasses impregnation with olive mill wastewaters has significantly accelerated the kinetic degradation of the tested biomasses, and has substantially increased the C2H6 yields production by the almond shell material o The application of impregnated biochars as amendment for sandy soil showed a significant increase in its water retention capacity, and also provided plants with significant amounts of nutrients, including potassium

× The assignment of the internship focused on the estimation of crop water needs using CROPWAT software

× This internship allowed me to discover the different components of the supply and drinking water System network