Laique Merlin Djeutchouang

Research and Support Scientist – Machine Learning, Ocean Systems & Climate

My work focused on updating and maintening a two-step machine learning approach based on CSIR-ML6 (Gregor et al., 2019) which combining the K-Means clustering and a three-member ensemble regression method made up of Support Vector Machine, Gradient Boosting Machine, and Feed-forward Neural Network algorithms. So far, I have:

• Perform code optimization and parallelism support.
• Analyzed about 30 GB of climate data to update the CSIR-ML6 approach, lowering the root mean square error by about 15%.
• Applied the resulting model to satellite and reanalysis data to predict the global CO₂ fluxes at the interface between ocean and atmosphere (see CSIR-ML-v2021), which was used for the Global Carbon Budget 2021 launched at The 2021 United Nations Climate Change Conference (COP26).
• Organised workshops on innovative programming frameworks such as Dask for processing and analyzing scientific Big Data by coupling Python and High Performance Computing (HPC) to speed up the analysis.
• Worked on large volume (about 5 TB) of multi-dimensional climate model data using multi-node systems with data-parallel processing and visualization using Dask and HPC systems.

Research Residency: Earth Science AI Engineer

My work focused on combining the use of AI with remote sensing sciences and satellite imagery data to tackle CH₄ emission problems from oil & gas extration sites including North Africa (Algeria), Permian Basin (North America), Turkmenistan and China. Basically, I:

• Researched, tested & developed machine learning-based plume masking algorithms for use in methane detection & monitoring pipeline.
• Implemented a land cover classification module for Sentinel-2 imagery data and integrated into Orbio Earth pipeline.
• Reviewed codes of other contractors & employees.
• Helped deliver pilot and customer satellite-based data.

Data Science Consultant

My work focused on assisting the director within the academic program, training and organizing outreach activities in STEM, but not limited to. Basically, I:

• Analyzed and scored 1000s of CVs/Resumes based on job specs; wrote programs (Python) to scrape and score LinkedIn & GitHub profiles, used in building of an AI-based recruitment system.
• Wrote a program (Python) to identify and assess individuals' Twitter footprint that is being used in the FAROSIAN social media screening.

Teaching Assistant (TA) & Head of TAs

My work focused on assisting the director within the academic program, training and organizing outreach activities in STEM, but not limited to. Basically, I:

• Organized lectures and lecturer evaluations by students and tutors.
• Assigned tutors to lectures, coordinated their work and reported to the academic director.
• Organized and led group/individual tutorials to improve STEM and programming (Python, R, MATLAB) knowledge and understanding, team-working and problem-solving skills of over 50 postgraduate students.

Research intern in Statistics & Data Science

My work focused on analyzing 12 GB of induced earthquake magnitude data. I used the Fisher-Tippett distribution to estimate the largest possible anthropogenic earthquake magnitude likely to be observed during multiple fluid injection operations (R).

Research intern in Machine Learning

My work focused on designing a Machine Learning framework to address survival modelling problems as predictive modelling problems (R).

Lecturing

I have done a few guest lectures and workshops at the University of Cape Town. In these workshops, students are taught to:

• Use Python IDEs such as Jupyter Notebook & JupyterLab, and Spyder, altogether with data analytics frameworks such as Numpy, Scipy, Pandas, Matplotlib and Xarray.
• Perform geospatial data processing and visualisation.
• Work with different ocean data sources and how to compare and plot these data.
• Leverage Dask framework for parallel processing and visualization of scientific Big Data.

Data Science – Ocean Systems and Climate, Department of Oceanography
Supervisors: Pedro Monteiro, Marcello Vichi
Title: Sampling scale sensitivities in surface ocean pCO₂ reconstructions in the Southern Ocean

Mathematical Sciences
By course work and dissertation and passed with distinction
Supervisors: Abdel Hameed El-Shaarawi
Title: Extreme value modelling with application to environmental control

Statistics, National Advanced School of Engineering
By course work and dissertation and passed with distinction
Supervisors: Henry Gwet, Eugene-Patrice Ndong Nguema, Wilson Toussile
Title: Modelling of the effect of intestinal bacterial flora on the vector ability of anopheles gambiae for resisting to plasmodium falciparum parasites

Department of Mathematics
Majoring in Statistics, Probability, Stochastic Processes & Modelling, Numerical Analysis, Algebra and Calculus.
Passed Honours in 2011 with upper second class.

This certification is awarded by Coursera. This certification can be verified here.

This certification is awarded by NVIDIA Deep Learning Institute and can be verified here.