AI Research Engineer with 8+ years in AI Safety & Enterprise Systems
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI safety researcher and systems engineer with demonstrated impact across clinical AI deployment, mechanistic interpretability, and AI governance. A Senior systems engineer with over 7 years in designing and developing Enterprise systems that have over 10,000+ users worldwide.
Makerere University
MSc Computer Science (AI Track) · Computer Science (AI Track)
August 1, 2022 – June 30, 2024
Makerere University
BSc Software Engineering · Software Engineering
August 1, 2014 – June 30, 2018
Google DeepMind
Scholar
January 1, 2022 – September 1, 2024
London, England, United Kingdom
Makerere University
Senior Systems Engineer
January 1, 2021 – Present
Kampala, Kampala District, Uganda
Emaisha Ltd
Lead AI Developer
December 1, 2018 – December 1, 2020
Kampala, Kampala District, Uganda
Thinvoid Ltd
Software Engineer
August 1, 2018 – December 1, 2018
Kampala, Kampala District, Uganda
WIMEA-ICT
Software Developer
June 1, 2017 – June 1, 2018
Kampala, Kampala District, Uganda
AI Safety: Mechanistic Interpretability in Clinical AI
January 1, 2026 – Present
Identified a spurious attention circuit (L8H10) in a Vision Transformer trained on chest X-rays that is causally verified as harmful yet completely invisible to standard output-level safety evaluation, with direct implications for clinical AI procurement and governance in low-resource settings. Finding replicates across architectures and datasets. Manuscript submitted to Deep Learning Indaba 2026 / IJCAI.
ITU Project: University of Oxford
February 1, 2025 – May 1, 2025
Collaborated on country-focused policy case studies (primarily Kenya), analyzing national responses to emerging AI safety institutes and governance structures. (Funded by Institute of Applied Policy)
Pluralistic Alignment: Annotation Bias & Hate Speech
January 1, 2025 – Present
Demonstrated empirically that majority-vote annotation in hate speech datasets suppresses minority value judgments at the hate/offensive boundary, a structural failure with direct implications for multilingual, multicultural populations across East Africa. Paper Accepted and published at ICML conference under Pluralistic Alignment Workshop @ ICML 2026.
AMR Surveillance AI Model
October 1, 2024 – Present
Served as Tech Lead in designing and piloting a low-resource recommendation models and companion mobile application for antimicrobial resistance (AMR) surveillance, co-infection detection, and real-time information dissemination to community. (Funded By: GIZ). Engineered efficient models optimized for low-compute environments to enable accurate monitoring and identification of AMR threats in data.
Tuberculosis Classifier and report Generation AI
January 1, 2023 – September 1, 2023
Directed creation of an explainable multi-task deep learning architecture for chest X-ray classification and automated report generation, overcoming long-context limitations in large language models for clinical usability. (Funded by Lacuna Fund). Built an interpretable AI system that streamlines TB screening workflows, with potential to accelerate diagnosis in high-burden settings.
Sickle Cell Early Detection AI
June 1, 2022 – Present
Led development of a lightweight deep learning model for enhanced, accurate interpretation of rapid diagnostic tests in low-resource contexts, improving early sickle cell disease detection. (Funded By: National Institutes of Health grant). Optimized computer vision algorithms to boost diagnostic reliability on mobile devices, addressing accessibility challenges in underserved regions.
When Majority Vote Silences Minority Values: Pluralistic Alignment Failures in Hate Speech Detection
Pluralistic Alignment Workshop @ ICML 2026
January 1, 2026 – Present
Spurious Circuits Invisible to Behavioural Evaluation: Mechanistic Interpretability in Clinical Vision Transformers
Deep Learning Indaba 2026 / IJCAI Journal
January 1, 2025 – Present
A Multi-Task Deep Learning Model for Classification and Report Generation from Chest X-Ray Images
Unknown
January 1, 2024 – Present
Minimal Idle Listening and Centralized Scheduling in TSCH Wireless Sensor Networks
IEEE, Greece
January 1, 2018 – Present
Condition Monitoring and Reporting Framework for WSN-based Automatic Weather Stations
IEEE, Botswana
January 1, 2018 – Present
Cultural Fit Analysis
The candidate's extensive work on projects addressing critical issues in East Africa (e.g., healthcare, agriculture, education, hate speech detection) and their involvement in AI governance for the region demonstrate a strong commitment to impactful, ethically-minded AI development. Their experience with diverse funding bodies (NIH, Lacuna Fund, GIZ, UNCDF) and collaborations with institutions like the University of Oxford and Google DeepMind indicate adaptability and a global perspective. This aligns well with organizations seeking engineers who are not only technically proficient but also socially conscious and capable of navigating complex, multicultural environments.
Soft Skills & Operational Fit
The candidate demonstrates strong leadership, problem-solving, and project management skills through their roles as Tech Lead and Lead AI Developer. Their involvement in policy case studies and governance structures indicates an ability to collaborate across diverse stakeholders and communicate complex technical concepts in a broader context. The focus on low-resource settings and accessibility suggests a practical, impact-driven approach to engineering.