Bernard Abuto
Machine Learning Research Engineer
I build the systems that bring machine learning research into practice.
My work sits at the intersection of machine learning research, systems engineering, and African language technology. I'm interested in what it takes to turn research ideas into reproducible experiments, reliable infrastructure, and useful AI systems.

A little about how I work.
I'm interested in the space between an idea in a research paper and a system that works in practice — and in being honest about what the results show.
Work at the intersection of research and systems.
A company I lead, two applied AI systems, and an ongoing research project — each a different part of the same question: how technical ideas become systems that can be understood, evaluated, and used.
- Sauti LabsAn AI research and products company building the models and technologies that let machines understand and generate speech across Africa’s linguistic diversity.
- Scribe AIAn offline-first church production system combining a local speech-intelligence prototype with a separately developed desktop application architecture.
- RentraA rental operations platform modelling the relationships, records, and workflows connecting tenants, landlords, properties, and rental activity.
- QueleaGuardInvestigating environmental, spatial, temporal, and agricultural drivers of red-billed quelea occurrence around Ahero Rice Irrigation Scheme.
Work recognised beyond the build.
Selected recognition for collaborative work in applied machine learning and AI.
FemScan
Chuka University AI for Health Innovation Challenge
Writing from the work.
Engineering decisions, research problems, and the convictions behind them — written from the projects themselves, including the parts that went wrong.
Spatial cross-validation when the data won’t spread out
Random k-fold overstates performance on clustered ecological data. The obvious alternatives broke on my dataset too — here is the fold design QueleaGuard ended up with, and why.
Read the postThe API chose my sampling design
How an undocumented sort order in a public biodiversity API put 97% of my background points in January — and what I changed so that kind of bias gets caught before a model ever sees it.
Read the postWhy I started Sauti Labs
It began with a speech model that didn’t understand the people I was building for. What I found when I went looking for alternatives, and why I think the distance between African-language research and usable products is worth building a company around.
Read the post