delphine nyaboke
Machine Learning & Software Engineer
Building responsible, inclusive AI for low-resource contexts
About Me
Machine Learning and Software Engineer with 4+ years of experience in designing, developing, and deploying AI/ML solutions, including LLM-based systems for data classification and automation in fintech and government applications.
Skilled in leading AI tool integration for evidence search/retrieval, synthesis workflows, and policy-facing platforms in low-resource African settings, translating business problems into scalable use cases like multilingual conversational AI serving 50M+ users, using Agile and cross-functional collaboration.
Proficient in prototyping, training, validating, and optimizing AI/ML models (e.g., deep learning, NLP, generative AI) with frameworks like PyTorch, Rasa, and OpenAI, while ensuring ethical AI principles, data privacy, and model governance through documentation, testing, and integration into production environments on cloud platforms (Azure, AWS, GCP).
Excel in ambiguous, distributed teams, calm in chaos; driving process improvements, data analysis, and scalable deployments to meet targets, with a commitment to inclusive, unbiased AI for sustainable development in multicultural contexts, with a focus on bias mitigation and ethical AI governance in low-resource African contexts.
Experience
AI & Machine Learning Specialist
International Centre for Evaluation and Development (ICED)
February 2026 – April 2026
Led the design and integration of AI-enabled tools to support evidence synthesis, systematic reviews, and policy decision-making.
- Built production-grade conversational RAG chatbot (LangGraph + Streamlit) for policy outputs from 500+ PDF collections
- Developed Flask file download API with SHA-256 integrity verification and secure streaming
- Created Django-based finance automation system integrated with Microsoft Graph API and Azure AD
- Designed RIS-to-PDF conversion pipelines with integrity checks
Machine Learning Intern
International Labour Organization (ILO)
June 2025 – December 2025
Developed and deployed an LLM-powered document classification and ranking tool that reduced processing time from 3 days to under 5 minutes.
- Built production LLM document classification system with human-in-the-loop validation
- Implemented CI/CD pipelines and ontology-based quality checks
- Tool remains actively used by the ILO Library
- Contributed to Swahili component of Global PIQA benchmark (arXiv:2511.03464)
Software Engineer
THiNK
2023 – 2025
Built multilingual conversational AI systems and automation tools for government and fintech platforms serving millions of users.
- Developed Rasa + Django multilingual chatbot integrated with Kenyan government platforms (eCitizen, etc.)
- Reduced operational costs by ~50% through LLM automation for bank/fintech processes
- Improved technical documentation coverage from 10% to 90%
Software Engineer
Sendy
2021 – 2023
Contributed to logistics and payments infrastructure across five African countries.
- Worked on tracking systems, M-Pesa integrations, and algorithm optimization
- Supported compliance and process improvements across multiple markets
Mozilla Fellow
2026–2027 Mozilla Fellow (Embedded)
Bias Lens
Co-designing and co-creating Bias Lens tool with journalists, content creators and fact checkers through hackathons and community workshops.
Piloting the open-source dashboard with over 100 journalists to support ethical, human-centered AI adoption in newsrooms and civic spaces across low-resource contexts.
- Selected as one of 10 global 2026–2027 Mozilla Fellows building alternatives to extractive tech systems
- Focus on responsible AI tooling that surfaces cultural bias and misinformation in AI-generated African stories
- Community co-design approach via hackathons with journalists, content creators and, fact checkers (in collaboration with Baraza Media Lab and Mozilla Foundation)
delphine nyaboke - Mozilla Fellow
2026-2027 Mozilla Fellows
Featured Projects
Global PIQA (Swahili Component)
Contributed to the Swahili portion of the Global PIQA benchmark for physical commonsense reasoning in low-resource languages.
Enhancing Modified VGG16 for Monkeypox Detection
Improved model for monkeypox detection using modified VGG16 and explored architectures like ConvNext for enhanced accuracy in data processing
Forecasting Electoral Violence by Leveraging Conflict Datasets
Integrated datasets for ML-based forecasting, involving data processing and analysis at election-year-country level.
AI-Powered Question Answering for Low-Resourced Schools
Developed conversational chatbot using Rasa, OpenAI, and LLMs for educational information management.
LLM-Powered Hospital Navigation System
Collaborated on LLM system to streamline patient appointment booking, focusing on data processing and integration.
Leveraging LLMs for Low-Resourced Languages
Built interface for science education in Kinyarwanda using LLMs, supporting information management in low-resource settings.
Personal
A few moments that ground me ...
With kids
With daughter and partner
Partner with kids
Partner
Me
Skills
Technical
- Python
- C/C++
- Java
- PyTorch
- LangChain / LangGraph
- Django
- Flask
- Docker
- Git
AI / ML
- Deep Learning
- Machine Learning
- Artificial Intelligence
- RAG Systems
- LLM Evaluation
- Multilingual AI
- Responsible AI
- Explainable AI
Languages
- Swahili (Native)
- English (Fluent)
- Mandarin(Basic)
- French(Basic)
- Abagusii (Native)
- Agikuyu(Basic)
Education
BSc Electrical and Electronic Engineering
Jomo Kenyatta University of Agriculture and Technology (JKUAT)2020
Get in Touch
Feel free to reach out for collaborations, speaking, or just to say hello.