Harouna TRAORE

- Automated trade data processing pipelines (Python, R, Dagster): quality control, cleaning, standardization and production of international trade indicators - Migrated SAS codebase to Python to industrialize analytical processes; documented business rules and processes - Built a SQL Server data warehouse for data storage, consolidation and reporting - Developed monitoring dashboards and data quality indicators (Power BI, R/Shiny, SAP Analytics Cloud) - Produced analytical reports on trade flows for institutional partners - Delivered targeted training on data tools, dashboards and analytics to partners - Developed and used AI tools for product classification, code generation and data querying

Trade Policy Analysis & Research - Contributed to Chapter 3 of "Trade for Peace: Pathways to Sustainable Trade and Peace" (WTO, 2025): integrated trade and peace datasets (IEP, Global Peace Index), applied econometric time-series models and correlation analyses on WTO membership and peace dynamics - Contributed to the DG Annual Report on Accessions: drafted the trade policy and market-access analysis section for Article XII countries and proposed new data visualizations to strengthen the evidence base - Developed comprehensive understanding of WTO accession processes: historical background, procedural steps (application, negotiations, ratification) and policy implications for acceding countries Trade & Development - Participated in the final Working Party meetings for the accession of Comoros and Timor-Leste: prepared meeting materials and drafted reports - Analysed key international trade issues: tariffs, non-tariff barriers, trade agreements and trade disputes at national and international levels Data & Statistics - Collected, cleaned and analysed trade, conflict and peace datasets from multiple international sources (IEP, WTO, UN) - Applied econometric methods (time-series models, correlation analyses) to quantify the relationship between WTO membership and peace indicators - Collaborated with researchers from the Institute for Economics and Peace to refine variable selection and ensure analytical robustness

- Backtesting of IFRS9 PD model: PiT vs ODR comparison, migration matrices, conservatism evaluation and recommendations - Built an R/Shiny application automating backtesting report production - reduced turnaround from several days to 15 minutes - Enriched performance indicators: PSI, HHI, AR/AUC, Hosmer-Lemeshow, conformity test, MAE/RMSE - Extended data quality diagnostics: missing values, outlier detection (Grubbs test), population stability - Conducted climate stress test analysis under NGFS scenarios (transition and physical risks) - Presented findings and recommendations to MRM teams and senior managers

Trade Data & Quality - Contributed to advise government officials from 5 African countries on trade data quality standards, governance frameworks and information-sharing protocols during African Trade Observatory (ATO) workshops - Prepared and delivered data quality reports for ATO partner countries, identifying inconsistencies and recommending corrective actions - Enhanced firm-level and aggregate trade data used for regional competitiveness monitoring across African markets Data & Systems - Upgraded the firm classification system (business type classification) to improve data categorisation and standardisation across ATO datasets - Contributed to the development of the African Trade Digital Interconnectivity (AfTDI) application and related technical documentation - Supported the operationalisation of data sharing protocols between national trade agencies and the ATO platform

Project: MARS (Modelisation, Apprentissage et Resilience des Socioecosystemes) - Designed a stochastic multi-compartment epidemic model (SEIRDS) as a data simulator for training and evaluation - Implemented and optimised Echo State Networks (Reservoir Computing) for epidemic time series prediction on simulated data - Evaluated model performance and sensitivity: RMSE analysis, prediction horizon sensitivity, robustness testing - Applied Deep Learning methods to a complex dynamical systems problem under supervision of CIRAD/IRD researchers