Solomon Nsumba
AI Researcher · Computer Scientist · Software Engineer
I'm a Computer Science PhD candidate at Makerere University, where I also serve as Technical Lead at the Makerere AI Health Lab. I work on applied AI and computer vision, mostly aimed at problems in healthcare and other resource-constrained settings.
As a Software Engineer at Sunbird AI, I also work on machine translation and language technology for Ugandan languages, and on applied AI for public health and urban sensing.
Research
My PhD research, based at Makerere University, focuses on using computer vision and wearable sensing to catch medication-administration errors before they reach a patient — work developed with clinical and engineering collaborators, including at the University of Washington. More broadly, my work spans:
- AI for Healthcare
- Computer vision and wearable sensing systems that support clinical decision-making and patient safety, including AI-enabled detection of medication-administration errors and smartphone-based diagnostic microscopy for resource-limited clinics.
- Computer Vision
- Visual recognition systems built for messy, real-world conditions — from crop-disease detection in smallholder farmers' fields to low-light image enhancement and medical image classification.
- African Language Technology
- Machine translation and language datasets for Ugandan languages, aimed at making AI systems and language tools usable for communities that mainstream models underserve.
- Applied AI for Public Health & Urban Sensing
- Large-scale data collection and analysis for public-sector and public-health applications, including one of the first detailed acoustic characterizations of a developing-world city.
Selected Work
AI-enabled wearable camera for medication-error detection
A head-mounted-camera system, developed with researchers at the University of Washington, that uses computer vision to flag drug-preparation errors — such as a vial swap — before they reach a patient. Evaluated across multiple operating rooms and hospitals.
Ocular — AI-powered diagnostic microscopy
A smartphone-to-microscope adapter paired with an AI diagnostic pipeline for malaria, tuberculosis, and cervical-cancer screening, built to cut slide-review time for healthcare workers in resource-limited settings. I lead the technical work on this project at the Makerere AI Health Lab.
SALT-31 machine translation benchmark
A discourse-aware, culturally grounded machine-translation benchmark covering 31 Ugandan languages across the Bantu, Nilotic, and Central Sudanic families, built to evaluate translation quality on realistic communication scenarios rather than isolated sentences.
Sunflower — a multilingual LLM for Ugandan languages
Open-source language models, built on top of Qwen 3, aimed at extending strong language understanding to Ugandan languages that mainstream LLMs handle poorly.
Urban-Noise-Uganda-61K
A dataset of over 61,000 annotated ambient-noise recordings and a noise map of Kampala and Entebbe — one of the first detailed acoustic characterizations of a city in a low-resource setting.
Computer vision for cassava disease and pest surveillance
Early work at the Makerere AI Lab applying computer vision to help smallholder farmers and researchers detect and quantify cassava disease and pest damage in the field, including two openly released image datasets.
Selected Publications
SALT-31: A Machine Translation Benchmark Dataset for 31 Ugandan Languages
Solomon Nsumba, Benjamin Akera, Evelyn Nafula Ouma, Medadi Ssentanda, Deo Kawalya, Engineer Bainomugisha, Ernest Tonny Mwebaze, John Quinn
Proceedings of the 7th Workshop on African NLP (AfricaNLP), 2026
Noise mapping and ambient sound recordings of the urban environment in Uganda
Solomon Nsumba, Tibabwetiza Muhanguzi, Evelyn Nafula Ouma, Imran Sekalala, Engineer Bainomugisha, Ernest Mwebaze, John Quinn
Scientific Data, 2026
Multimodal Approach for Cervical Cancer Histopathology Classification and Automated Diagnostic Report Generation
Rose Nakasi, Cosmas Wamozo, Solomon Nsumba, Benjamin Rukundo, Tonny Okecha, Byron Mubiru
CVPR Workshops (CVMI), 2026
Sunflower: A New Approach To Expanding Coverage of African Languages in Large Language Models
Benjamin Akera, Evelyn Nafula Ouma, Godfrey Yiga, Peter Walukagga, Phionah Natukunda, Tobius Saaka, Solomon Nsumba, Lilian Teddy Nabukeera, Jonathan Muhanguzi, Imran Sekalala, Naome J. Namara, Engineer Bainomugisha, Ernest Mwebaze, John Quinn
arXiv preprint, 2025
Detecting clinical medication errors with AI enabled wearable cameras
Justin Chan, Solomon Nsumba, Mitchell Wortsman, et al.
npj Digital Medicine, 2024
Machine translation for African languages: Community creation of datasets and models in Uganda
Benjamin Akera, Jonathan Mukiibi, Lydia Sanyu Naggayi, Claire Babirye, Isaac Owomugisha, Solomon Nsumba, Joyce Nakatumba-Nabende, Engineer Bainomugisha, Ernest Mwebaze, John Quinn
3rd Workshop on African Natural Language Processing, 2022
Experience
Current
- PhD Candidate, Computer Science · Makerere University
- Technical Lead, Machine Learning & Computer Vision · Makerere AI Health Lab
- Software Engineer · Sunbird AI
Previous
- Research Software Engineer · Makerere AI Lab
Education
- PhD, Computer Science · Makerere UniversityIn progress
- MSc, Data Communications and Software Engineering · Makerere University
- BSc, Software Engineering · Makerere University
Talks & Community
- Panelist & Masterclass Speaker, "Digital Health for All: Transforming Healthcare Access and Outcomes through Innovation" — AI Summit, Speke Resort Munyonyo, 2025
- Presenter, "AI & Microscopy: Making Diagnosis more Accurate and Faster" — AI in Health Africa Conference, Kampala, 2025
- Presented the Ocular project's diagnostic AI work — AI in African Health Conference, Kampala, 2024 Details
- Panelist, "Harnessing the power of AI: building stronger communities through interdisciplinary collaboration" — Deep Learning IndabaX Uganda, 2023 Details
- Coordinator, Weekly Artificial Intelligence Seminars — Makerere AI Lab, 2015–2020
- Poster Presentation (Most Outstanding Poster Award), "Scaled deployment of a crowdsourcing system for determining the health of cassava in small-holder farmer fields" — COMPASS, 2019
About
I got into this work through software engineering, building systems before turning to the research questions underneath them. That’s shaped how I approach AI research: I’m most interested in models and systems that hold up outside a benchmark — in a clinic, a farmer’s field, or a conversation in a language most tools don’t support.
That’s the thread connecting my work across computer vision, digital health, and language technology: building and evaluating AI systems for the conditions they’ll actually run in, rather than assuming the conditions of a lab.