Experience

Clinityx
• France
• IT Services and IT Consulting
• 1 - 100 Employee
• Data Scientist

  • Feb 2023 - Present


Aive

• Pakistan
• Wholesale Building Materials
• 1 - 100 Employee

• Data Scientist intern

  • May 2022 - Sep 2022

  Trained an object detection model (based on SOTA Yolov5) to detect logos in videos. The best model reached 0.79 in recall, 0.86 in precision and 0.86 in mAP. • Setup and deployed Label Studio on EC2 instance for managing data augmentation remotely. • Researched best metric to evaluate the models’ performances for which false positives are much more dire. • Performed data analysis and tracked false positives post-model training to better understand model results (i.e. why are there false positives, why does the model perofrm better on certain brands than others). • Optimised code for production: Migration from CPU to GPU, Optimisation of YOLOv5’s detections by adding multi-object trackers, Add acceptance tests and unit tests.. • Presented research paper results for SOTA Single Object Tracking using Transformers. Show less



Ministry of Economy and Finances

• Mozambique
• Government Administration
• 1 - 100 Employee

• Data Scientist | Data Analyst

  • Mar 2021 - Jun 2021

  Household debt analysis in Morocco • Apply VAR and ARDL time series models in order to capture the relationships between the household credits and the other economic sectors in the short and long long term. • Simulate shocks on the different variables. • Predict the level of houshold credit with deep models Learning like the LSTM. Household debt analysis in Morocco • Apply VAR and ARDL time series models in order to capture the relationships between the household credits and the other economic sectors in the short and long long term. • Simulate shocks on the different variables. • Predict the level of houshold credit with deep models Learning like the LSTM.



National Institute of Statistics ans Applied Economics (INSEA)

• Data Scientist | Data Analyst

  • Jul 2020 - Aug 2020

  Moroccan Air traffic analysis. • Descriptive analysis of air traffic in Morocco. • Air traffic forecasting using exponential smoothing and Box-JenKins ARIMA family meth- ods. (Eviews, Rstudio) • Air traffic time series forecasting using LSTM model. (Python) Moroccan Air traffic analysis. • Descriptive analysis of air traffic in Morocco. • Air traffic forecasting using exponential smoothing and Box-JenKins ARIMA family meth- ods. (Eviews, Rstudio) • Air traffic time series forecasting using LSTM model. (Python)


the Moroccan Retirement Fund (CMR)
• Data Analyst

  • Jul 2018 - Jul 2018