CV/Mission Pandas freelance

Exemple des missions de Moukouba,
freelance PANDAS résidant dans la Gironde (33)

• Overstory Case Study
  2023 - aujourd'hui

  Cloud segmentation using Sentinel-2 satellite images
   Method: Implementation of a Deep Unet-like model with Efficient Inception block for Semantic Segmentation;

  Python, Pandas, OpenCV, Sklearn, Rasterio, TensorFlow/Keras.

• VEOLIA Challenge
  2022 - aujourd'hui

  Predicting odor compound concentrations (Air SO2 concentration forecasting)
   Method: Hybrid CNN-GRU Model for Parallel Multiple Inputs Multivariate – Multi-step time series forecasting.
   Tools: Python, Pandas, TensorFlow, Keras.

• Expleo Group/Flying Whales, Cross
  Jan 2021 - Jan 2022

  Department Meteorology and Data Science Referent
   Conducted data extraction and assessed outputs from Numerical Weather Prediction models and carried out geospatial data
  analysis and mapping on Copernicus datasets.
   Built a cloud based and on-premise terabyte-scale data warehouse.
   Developed and optimized a decision-support software tool (Python-based) to facilitate operations, mission planning,
  scheduling, and real-time assistance, using parallelization techniques for CPU and GPU to improve computational efficiency.
   Served as an in-house expert, providing assistance across all departments in design activities, and mission planning and
  definition for airship construction and operation.
   Undertook climatological studies, mined data, and developed machine learning models for climate/weather prediction in
  targeted areas and anchor sites for airship operations.

• Schlumberger Challenge
  2021 - aujourd'hui

  Sinusoid segmentation in subsurface images
   Method: Implementation of a Deep Unet-like model with nested Inception blocks for Semantic Segmentation;

  Python, Pandas, OpenCV, Sklearn, Skimage, TensorFlow, Keras.

• Kaggle Challenge
  2020 - aujourd'hui

  Global Land Average temperature time series forecasting (see code on github).
   Method: Implementation of a GRU architecture (Gated Recurrent Units), with hyperparameter optimization.

  Python, Pandas, Sklearn, Plotly, TensorFlow, Keras, Talos.

• Data Scientist

  ANAC/National Dept. of Meteorology, Congo Climate
  Jan 2015 - Jan 2018

  Developed operational tools in Python for statistical climate modeling and forecasting, with applications across sectors such as
  agriculture, rural community activities, health, energy, and construction.
   Performed geospatial data analysis and mapping on various GIS platforms.
   Performed environmental and health risk analyses, employing Bayesian Probabilistic modeling.
   Extracted, preprocessed, and derived new local variables via feature engineering on heterogeneous data obtained from various
  sources, including in-situ sensors, satellites, and reanalysis.
   Developed Machine Learning models for climate forecasting, and Computer Vision systems for earth observation.
   Maintained, fine-tuned, and scaled models; automated processes and managed the national climatic database.
   Regularly communicated findings, attended global conferences and workshops, shared knowledge through internal training
  sessions, and mentored interns.

• Dept. of Mathematics of the Autonomous University of Barcelona, Intern
  Jan 2015 - Jan 2015

  Received mentorship in statistical modeling and operational research methodologies.
   Conducted research on Machine Learning algorithms, focusing on Low-Rank Matrix Factorization for Recommender Systems,
  large Matrix Completion, as well as data smoothing and filtering.
   Implemented, from the ground up, algorithms including Maximum Likelihood PCA, Singular Value Thresholding, and Soft-Impute Alternating Least Squares as integral components of Master's Thesis research.

• Meteorologist

  ANAC/National Dept. of Meteorology, Congo
  Jan 2012 - Jan 2014

  Developed operational tools in R for weather and seasonal climate forecasting as well as air pollution dispersion prediction.
   Conducted river flow modeling and flood forecasting; analyzed associated risks and impacts across various socioeconomic
  sectors.
   Designed agricultural experiments and forecasted crop yields using climatic data.
   Coordinated a national project on air quality monitoring and development of biometeorological products.
   Led the task force on national climate watch, managed the national climatic database, and produced seasonal climate outlooks
  for a variety of end-users such as the general public, agriculture sector, health care, energy providers, transportation, and the
  construction industry.
   Ensured effective dissemination of information to both technical and non-technical end users, while enforcing the WMO
  Quality Management System.
   Represented the country as needed, with various external entities, such as acting as Focal Point for the ICAO – EU Carbon
  Offsetting and Reduction Scheme for International Aviation (CORSIA).

• Cloud Data Engineer

  CGI, Bordeaux, France,
  aujourd'hui

  Engineered and monitored data pipelines utilizing Azure cloud.
   Developed Machine Learning solutions for the prediction of water resources demand in the Bordeaux metropolitan area.
   Developed Object Detection and Tracking systems based on YOLO architectures.
   Created, updated, and maintained Machine Learning pipelines, for various applications, using MLOps tools.
   Developed, deployed, and managed web applications along with back-end Data/Machine Learning APIs.
   Ensured the scalability and reliability of cloud-based data systems.

• EDF R&D Challenge
  aujourd'hui

  Individual household electric power consumption forecasting
   Method: Implementation of a Transformer architecture for Multivariate – Multi-step time series forecasting.
   Tools: Python, Pandas, TensorFlow, Keras.