Professional Objective
I am seeking a research opportunity at the intersection of AI safety, global catastrophic risks, and sustainable development. I aim to apply my skills in machine learning, computer vision, and scientific computing to investigate high-impact problems in biosecurity, climate resilience, and robust AI systems.
Education
African Institute for Mathematical Sciences (AIMS)
AI for Science Master's Programme (Funded by Google DeepMind)
January 2025 - November 2025 (Expected) | Cape Town, South Africa
University of Yaounde I
Master degree in Computer Science (Data Science)
2019 - 2024 | Yaounde, Cameroon
Honors and Awards
- Google DeepMind Scholarship (2025) - For the AI for Science programme at AIMS.
- African Health Award (2023) - Third Prize at the Hackathon with Intelligent Patient Administration System (IPAS), Digital Health Hackathon.
Research Experience and Publications
Master Thesis – Robust Deep Learning for Climate Forecasting
AIMS South Africa (Ongoing) | Supervisor: Dr. Shruti Nath
- Designed and tested ML architectures under non-stationary data conditions.
- Surveyed climate modeling approaches and uncertainty quantification techniques.
- Developed methods for multi-source and multi-scale data integration.
- Applied neural networks for complex time series analysis.
Publication: Diagnosis of Plant Diseases Using Mobilenet and Inception Architectures
SSRN, 2024.
- Developed classification models to identify plant diseases from images.
- Applied transfer learning techniques to maximize performance with limited data.
Key Projects
Project A: Credit Risk Modeling with Machine Learning
Built end-to-end machine learning models to predict loan default probabilities. Feature engineering, XGBoost tuning (AUC-ROC 0.89), SHAP for explainability.
Tools: Python, Scikit-learn, XGBoost, SHAP, Pandas, Matplotlib
Project B: Climate Simulation and Downscaling with Neural Networks
Predicted temperature and precipitation from CMIP6 global climate model outputs. NetCDF data preprocessing, bias/variance analysis, interpretable visualizations.
Tools: Python, TensorFlow, NetCDF4, CMIP6 Data, Scikit-learn, Matplotlib
Project C: Time Series Forecasting for Financial Market Volatility
Developed models (Facebook Prophet, LSTM) to forecast short-term SP 500 volatility. Evaluated using MAE, RMSE, and directional accuracy.
Tools: Python, Facebook Prophet, LSTM (Keras), Scikit-learn, Pandas, Plotly
Project D: OCR project
Designed a cloud-based Optical Character Recognition (OCR) system. Used Tesseract OCR with OpenCV, deployed with AWS Lambda, Terraform, Docker.
Tools: Python, Tesseract OCR, AWS Lambda, Terraform, Docker, OpenCV
Workshops and Scientific Outreach
I-X AI in Science
AIMS South Africa and Imperial College London (April 2025 - May 2025, Notebooks)
Interdisciplinary workshop on AI/ML application in scientific research (chemistry, biology, physics, mathematics). Core topics: Classification, clustering, deep learning. Hands-on Python labs.
Science and Engineering of Large Language Models (LLMs)
AIMS South Africa, Google DeepMind, MILA and OXFORD (March 2025 - April 2025, Notebooks)
Foundations of LLMs, training, optimization, advanced architectures (MoE), adaptation (LoRA), scalability (FSDP).
Modern Statistics and Machine Learning for Population Health in Africa
AIMS South Africa and Imperial College London (March 2025 - April 2025, Notebooks)
Implementation of probabilistic programming languages (Stan, NumPyro), Bayesian methods, modeling regional challenges (disease outbreaks, air pollution).
ATLAS CERN International Masterclass – Particle Physics Outreach
AIMS South Africa, IPPOG (March 2025)
Guided high school students in identifying particles using real data from CERN's ATLAS experiment.
Technical Skills
Programming Languages:
- Python
- R
- SQL
- C
- C++
- JavaScript
- Java
- ROS
- Cuda
Web Technologies:
- HTML
- CSS
- JavaScript
- React
- Flask
- Django
Database Systems:
- MySQL
- PostgreSQL
- MongoDB
Data Science & Machine Learning:
- Scikit-Learn
- TensorFlow
- Keras
- PyTorch
- Apache Spark
- JAX
- SHAP
- LIME
- OpenCV
Cloud Technologies:
- AWS
- Google Cloud Platform
- Microsoft Azure
DevOps & Version Control:
- Git
- Docker
- Jenkins
- Kubernetes
- Ansible
- Streamlit
Mathematical & Statistical Tools:
- MATLAB
- Excel
- Statistical Analysis System (SAS)
- SPSS
- Mathematica
Other Tools & Technologies:
- Tableau
- Power BI
- Jupyter Notebook
- Apache Spark
- Hadoop
- HYPATIA
- Prophet
Contact Me
Feel free to reach out for research opportunities, collaborations, or any other inquiries.
- Email: melvin@aims.ac.za
- Phone: +27 713 910 736
- GitHub: MELAI-1
- LinkedIn: astridemelvinfokamninyim
- Location: Muizenberg, 7945 Cape Town, South Africa