Youssef Makni

DSS Governance & Centralized Monitoring • Designed and implemented an advanced Python-based monitoring system leveraging the dataikuapi to inventory all DSS objects (projects, datasets, recipes, jobs, scenarios, metrics, checks). • Built and deployed daily interactive dashboards for automated quality control, anomaly detection, and best practices enforcement. • Enabled DSS Event Server for centralizing logs from both Design and Automation nodes. • Designed a monitoring for eks consumption based on logs with daily dashboards Integration & External Connectors • Implemented and managed key integrations: Dataiku ↔ AWS (S3, Athena via CDH), Databricks, Azure OpenAI (via LLM Mesh), SharePoint & Teradata connectivity Advanced Development (Python, Plugins, Macros) 1. Smart engine recommendation for recipe execution (Spark, in-database, DSS). 2. Smart compute of dataset records across flows using the most efficient engine (Spark, Hive, SQL, DSS). 3. Automated S3 file copy operations between buckets. 4. Custom data quality checks packaged as reusable "check"-type plugins. 5. Flow refactoring through Dataiku Applications. 6. Spark pipeline implementation with virtualization of intermediate datasets (execution time reduced by 60%). Training & Business Enablement • Led adoption initiatives across business units: Finance, Data Energy Management, and Go-to-Market teams. • Delivered tailored training sessions based on user profiles o Illustrated slide decks o Business-aligned hands-on use cases. o Curated best practices for enterprise-wide Dataiku adoption. Technical Env : Dataiku, Python, Spark,AWS (S3,Glue,Athena),CDH, SQL, Databricks, Teradata, Power BI.

. Conducted comprehensive training for a team of six members at BNP Leasing Solutions on Dataiku, covering all concepts from Core Designer, Advanced Designer, ML Practitioner, MLOps Practitioner, and Developer certifications. . Emphasized advanced coding practices in Dataiku, including leveraging the Dataiku Python package (classes and methods), developing Dash web apps, shared code : libraries and git, customized scenarios, plugins, SQL queries, scripts, and notebooks. . Guided the team in developing machine learning analyses on their datasets and facilitated the seamless migration of their projects to Dataiku.

• Streamlined data preparation workflows within Dataiku, leveraging SQL pipelines to significantly cut execution times and reduce load on Teradata and Oracle databases. • Developed customzied scenarios in Dataiku, ensuring private bankers receive personalized alerts and client-specific information. • Oversight of project deployment to production on Dataiku. • Development of a dedicated pipeline API for WealthMonitor to proactively identify potential clients. • Development of a dedicated pipeline API for Morningstar to monitor the daily evolution of stock portfolios for high-net-worth clients. • Design of a monitoring system for the Wealth Management Dataiku tenant in Python, utilizing the Dataiku API, for comprehensive visibility of projects, datasets, connections, jobs, recipes, and scenarios. • Optimization of data management through the use of a custom monitoring tool, enabling the consolidation of duplicated SQL tables into data collections and reducing database resource usage. • Innovation through the development of a flow branch analysis algorithm, including the automatic extraction of SQL table names from SQL queries, for smarter data management. • Developed a Python algorithm to automatically traverse active scenarios, detect anomalies, and suspend the execution of affected scenarios to maintain data process integrity. • Developed a Python algorithm to automatically traverse all datasets in Dataiku connected to SQL databases, calculate data skew, and optimize database storage by identifying and indexing primary key columns for highly skewed tables lacking primary indexes. • Technical Env: Dataiku, Python, SQL, Teradata, Oracle

• Participated in the development of multiple data science use cases across different sectors (banking, retail, sustainability... ) on dataiku. • Developed on dataiku the 'Anti Money Laundering triage' use case to help financial institutions detect fraudulent transactions and sort alerts for the compliance team • Gathered data from different sources, cleaned and developed webapps to visualize data, and designed machine learning algorithms on dataiku • Pitched use cases at various events (Everyday AI 2022, salon de la Data de Nantes 2022) • Technical Env : Dataiku, Python, Dash, Snowflake

Building tools to describe datasets and query knowledge graphs in natural language • Proposed an end-to-end tool to automatically extract heterogeneous metadata of a dataset from any given textual description. • Studied the revolutionary paper ’attention is all you need’ to apprehend state-of-the-art attention mechanisms, transformers, and neural machine translation. • Created a neural machine translation from scratch to automatically translate English queries to structured SPARQL queries using Pytorch in order to query knowledge graphs in natural language . • Coded a Transformer architecture that achieved a bleu score of 0.8 using DBPedia Data. • Technical Env : Python, Transformers, PyTorch, Knowledge Graphs, Sparql