Applied AI/ML Engineer with 20+ years as a software engineer. The last 13 were spent architecting and building OppiaMobile, an open-source, offline-first learning platform I founded, deployed to 30,000+ health workers across Ethiopia, Liberia, India, and Nigeria, with teams in Ghana and Pakistan. That included delivering a health-worker training platform from scratch, solo, in under four months to meet a hard USAID launch deadline.
I've recently submitted my MSc dissertation in AI at the University of Leeds: a framework and dashboard, XAILens, for comparing how classical machine learning models and LLMs explain their predictions in healthcare settings, evaluated through task-based usability testing with real users.
I'm now building on that with an AI-powered simulated-patient tool for clinical training. It uses separate patient and examiner LLM roles, each independently configured: the patient generates the conversation, and the examiner grades the resulting transcript against a rubric.
I'm now looking for my next role in applied AI/ML engineering. The common thread through my career has been real-world systems that have to actually work for real users under real constraints, and I'm applying that same discipline to ML and LLM systems now.
A framework to generate, visualise, and compare explanations from classical machine-learning models and LLMs, applied to healthcare prediction.
Data‑mining project analysing conversational patient-doctor text using interpretable ML and LLM‑based explanations for ICD‑10 classification.
Designed and deployed a containerised machine‑learning xApp for real‑time network traffic classification, evaluated under load using Kubernetes, Prometheus, and Grafana.
Conceived, initiated, and led the development of an open‑source, offline‑first mobile learning platform for healthcare training, deployed at national scale in collaboration with Ministries of Health.
Consultant to the University of Alcalá's international development programme, continuing OppiaMobile development that began at Digital Campus.
- Owned architecture and backend development on the Django server throughout, while a growing part-time developer team took on Android work as the platform evolved.
- Delivered technical development for mobile health deployments in Ethiopia, Liberia, and Nigeria, reaching over 30,000 health workers
- Open-source platform adopted and deployed in Ghana and Pakistan with no involvement from the project team.
- Ran requirements and delivery planning directly with NGO and Ministry of Health partners.
- Maintained Django and Android test suites in GitHub Actions CI, reaching 80%+ server-side coverage.
Part-time/contract role, running alongside Co-founder & Developer at Digital Campus (2011–2018) and, from 2018, alongside Technical Advisor / Consultant at the University of Alcalá.
- Engaged to manage partner content submissions for the ORB platform, an open repository of health-worker training resources.
- When the externally contracted build fell behind schedule, proposed and personally built the platform solo in 3–4 months to meet a fixed May 2014 launch deadline, avoiding a re-procurement cycle.
- Ran ongoing technical development and partner support.
- Co-founded a non-profit applying technology to healthcare education and training, conceiving and initiating OppiaMobile as a solution to close gaps in mobile learning for health workers.
- As the organisation's first developer, architected and built OppiaMobile v1 end-to-end (Django, MySQL, AWS, native Android app, Moodle plugin), later bringing on part-time developers from 2014 to scale delivery.
- Built the Django-based analytics dashboard for supervisors and project managers to track health workers' progress through courses.
- Secured UK Aid Direct (DFID) funding for a multi-year Ethiopia programme, accountable for technical delivery, reporting, and outcome evaluation.
- Managed organisational finances and operations as the only paid employee, coordinating with NGO partners to align technical delivery with operational constraints.
- Prototyped early concepts for what became OppiaMobile and led to co-founding Digital Campus and securing DFID funding for its first full implementation.
- Sole developer of a rules-based antenatal care dashboard (built on ODK, with a custom PHP layer), piloted with health workers in Ethiopia as part of a PhD programme to test smartphone-based patient management, an approach still largely unproven at the time. Findings published in PLoS One (2013).
- Placement with Voluntary Service Overseas (VSO) implementing e-learning systems and building IT/training capacity.
- Set up two computer labs using Sun Solaris thin-client architecture, for centralised management and lower maintenance overhead compared to a standard PC lab.
- Delivered elearning training to 50 lecturers across Medical and Engineering faculties.
- Developed social software tools, including instant messenger integrated into Moodle, for the OpenLearn platform.
- Built the "Knowledge Network" educational technology research and resource-sharing platform
- Built open-source ed-tech tools (a Learning Design player, JISC-funded projects) across ColdFusion, JSP/Servlets, and PHP.
- Included 2 month placement at Macquarie University, Sydney.
Selected modules: Machine Learning, Deep Learning, Data Mining & Text Analytics, Cloud Computing, Algorithms
- Thesis project: A Proof‑of‑Concept Framework for Comparing Machine Learning and Large Language Models in Healthcare Using Explainable AI
Mentored student teams and supported 150+ participants during the University of Leeds Computing Society annual hackathon.
Managed day‑to‑day operations of a co‑working and studio space for a community of 15 artists and makers in Joensuu, Finland.
International Journal of Medical Informatics (Shannon, Ireland), 101, 9–14. https://pubmed.ncbi.nlm.nih.gov/28347452/
This report shares an initial blueprint to create a scalable, locally sustainable, end-to-end content distribution process that uses mobile technology to provide frontline health workers access to relevant health content.
PloS One, 8(10), e77563. https://doi.org/10.1371/journal.pone.0077563