Hecham Abakouy

AI-Powered Anti-Plagiarism, Double-Funding & Multiple-Submission Detection Platform

Sole architect of an end-to-end AI platform helping National Agencies across Europe identify fraudulent or duplicate Erasmus+ project submissions —combining NLP pipelines, ML-based similarity scoring, LLM enrichment, and interactive decision-support dashboards.

▸ Architected the full data pipeline: ingestion and normalisation of large volumes of unstructured application texts from multiple sources using Python, Oracle SQL and ETL frameworks

▸ Implemented a multi-layered NLP similarity engine (TF-IDF, cosine similarity, sentence embeddings) capable of flagging plagiarism and duplicate submissions at scale across thousands of annual applications

▸ Integrated OpenAI GPT via REST API for semantic enrichment, contextual keyword extraction and AI-assisted narrative analysis; applied advanced prompt engineering to optimise precision and recall

▸ Designed data models and scoring layers for similarity ranking, deduplication and cross-year historical comparisons; ensured full data quality, reproducibility, traceability and auditability aligned with FAIR principles

▸ Built and maintained interactive dashboards and KPI reports in QlikSense, QlikView and Power BI, rolled out to national stakeholders across 30+ countries for programme decision-making

▸ Contributed to scalable data architecture aligned with Azure Synapse / Microsoft Fabric / Lakehouse principles; integrated external datasets via REST APIs and automated enrichment workflows

Designed and developed Python-based data quality and enrichment pipelines to consolidate research datasets across the H2020 programme, integrating heterogeneous external sources via APIs and web scraping.

▸ Consolidated H2020 participant company data and automated SME classification by integrating the Orbis database (Bureau van Dijk) via Python; reduced manual verification effort significantly

▸ Built REST API connectors and web scraping pipelines (BeautifulSoup, Selenium) to enrich IPR datasets from EPO, Google Patents, CrossRef and Scopus; validated and reconciled data across sources

▸ Implemented ML-based record linkage, deduplication and entity disambiguation using the Record Linkage Toolkit; clustered and reconciled research organisations across heterogeneous datasets

▸ Developed data quality validation rules and automated quality control checks, ensuring consistency and traceability across multiple integrated datasets

▸ Integrated cleaned and enriched datasets into downstream BI reporting pipelines; maintained technical documentation and contributed to team knowledge base on Confluence

Key contributor to the European Commission's financial data warehouse —covering reporting development, SQL optimisation, BO Universe management and QlikView dashboard delivery for the budget directorate.

▸ Analysed business requirements from financial users and translated them into data warehouse evolutions, new report specifications and SQL optimisation strategies

▸ Developed and optimised complex Oracle SQL queries and Business Objects reports (Webi & Rich Client) for 100+ financial reporting users across the directorate

▸ Created and maintained BO Universes, improving data layer abstraction and enabling greater self-service reporting adoption among non-technical users

▸ Managed the full migration of the reporting suite from BO DESKI 6.5 to BO XI, coordinating testing, user acceptance and rollout; verified data consistency across ABAC, SAP and the data warehouse

▸ Built QlikView dashboards for budget monitoring and financial decision-making support; defined KPIs in collaboration with business stakeholders

▸ Contributed to data warehouse change management processes driven by ABAC software updates; maintained quality assurance across all integrated financial data systems