We build
end-to-end
systems.
theNatives is a software engineering studio. We design, build, and deliver technical systems — from ML pipelines and commerce platforms to geospatial tooling and fintech infrastructure — for clients across East Africa, the region and beyond.
Full-stack systems from schema to interface. Monoliths, microservices, APIs.
Model training, inference pipelines, dataset curation, embedding models, computer vision.
Analytics infrastructure, ETL pipelines, geospatial data, satellite imagery processing.
Native Android (Kotlin/Jetpack Compose), React/TypeScript SPAs, design systems.
M-Pesa integrations, payment state machines, embedded financial infrastructure.
QGIS map servers, land cover segmentation from satellite imagery, QGIS plugin development.
Smart contracts, on-chain logic, decentralised application architecture.
Supply chain orchestration, event-sourced order lifecycle, cross-border commerce platforms.
Cross-Border Commerce Platform
A vertically integrated B2C import commerce platform with embedded logistics execution, native M-Pesa payments, and end-to-end order visibility across thirteen discrete states. The system owns every stage of the customer's goods from factory floor to hub pickup — procurement, consolidation, sea freight, customs clearance, last-mile, and settlement.
The tracking system is a multi-stage event-sourced order lifecycle manager. The payment layer is built natively around M-Pesa STK push — the payment rail and the order management system are a single cohesive state machine. The recommendation engine runs an Item2Vec embedding model over purchase and browsing history.
QGIS Map Server & Land Cover Segmentation
An interactive web interface on top of a QGIS map server allowing users to select layers and export high-quality PDF maps. A companion CNN model segments land cover from optical multiband satellite imagery, packaged as a QGIS plugin for field use.
- Item2Vec for Sparse Purchase Histories: Embedding Products in Implicit Feedback Regimes How we adapted Word2Vec's skip-gram architecture to learn product embeddings from sparse, implicit purchase signals — without collaborative filtering at scale.
- ResNet-18 for Image Classification: Residual Learning on MNIST and CIFAR-10 How we implemented a ResNet-18 classification system with dual CLI/GUI interfaces, and what the residual connection formulation buys you over plain convolutional stacks.
- Semantic Segmentation of Land Cover from Multispectral Imagery: A CNN Approach Our methodology for training a convolutional segmentation model to classify land cover types from optical multiband satellite imagery, and how we packaged it as a QGIS plugin.
Based in Nairobi. Working across East Africa and beyond.
hello@thenatives.dev