Projects
Research pipelines, engineering work and competition solutions. Code is linked where it is public; where it is not, the write-up describes the method in enough detail to reproduce it.
-
Multi-modal crop yield prediction — Cameroon
Six deep learning architectures over three nested datasets — satellite, climate, socio-economic — for maize, cassava and cocoa.
-
Google Earth Engine satellite extraction pipeline
Monthly NDVI, EVI, NDWI, NDRE and Sentinel-1 radar indices for Cameroonian crop regions, with a Landsat 5 to GIMMS fallback for pre-2000 gaps.
-
IoT health platform with embedded AI
A connected health platform with embedded AI for patient follow-up in sub-Saharan Africa, built for intermittent and incomplete data.
-
Reconstructing incomplete agricultural production series
Ensemble statistical reconstruction under hard agronomic constraints, to fill and backfill maize and onion production series for the Centre region.
-
BirdCLEF 2026 — acoustic species identification
EfficientNet-B0 sound event detection combined with Perch v2 embeddings. Macro ROC-AUC ≈ 0.930.
-
ROGII — wellbore geology prediction
TVT prediction via a LightGBM/XGBoost ensemble with a Bi-LSTM component, rebuilt for fast execution under submission constraints.
-
Shadow-based geolocation from synthetic imagery
GPS and UTC time from rendered scenes, without analytical solar solvers. A three-seed ResNet-34/ResNet-50 ensemble averaged in xyz and sin/cos space.
-
Knowledge tracing from tutoring transcripts
Predicting pupil correctness from tutoring session transcripts, evaluated on log loss where calibration matters as much as ranking.