My Recently Works
A collection of the various projects I've been privileged to work on to tackle challenges in various Industrial sectors.
The UTT AMIS Fund Analytics Platform is a comprehensive financial analytics solution designed to democratize access to investment data for Tanzanian mutual fund investors. Recognizing that fund performance information was often fragmented and difficult for everyday investors to interpret, I developed this platform to transform over five years of raw NAV data into meaningful, actionable insights. The platform covers all six UTT AMIS funds: Umoja, Wekeza Maisha, Watoto, Jikimu, Liquid, and Bond, providing users with institutional-grade analytics through an intuitive interface that works seamlessly across desktop and mobile devices.
From a technical perspective, the platform showcases expertise across the full data analytics and engineering spectrum. The front-end features interactive D3.js visualizations with smooth animations, responsive tooltips, and PNG export functionality for each chart. Core financial metrics are computed programmatically, including compound annual growth rates, annualized volatility, Sharpe and Sortino ratios, and maximum drawdown analysis. The platform also incorporates four distinct investment calculators, SIP projections, lump sum modeling, goal-based planning and historical scenario analysis, enabling users to make data-driven investment decisions. This project reflects my commitment to building solutions that combine technical rigor with real-world impact, making complex financial data accessible to those who need it most.
- Deployed, managed, and maintained Metabase BI platform across multiple departments including Finance, Operations, Risk, Sales and Customers Support
- Designed and developed interactive analytical dashboards providing real-time visibility into payment transactions, agent networks and customer activity
- Built data pipelines integrating Metabase with REST APIs, PostgreSQL, MySQL databases, and core banking systems
- Automated ETL processes with 15-minute refresh intervals, enabling near real-time reporting for daily operations
- Managed user access controls, query optimization, and infrastructure maintenance ensuring high availability for business-critical analytics
- Reduced manual reporting workload by 80% through self-service analytics and automated data workflows
Technologies: Metabase, Python, SQL, PostgreSQL. REST APIs, ETL, Cron Jobs, Linux, Docker
Developed an AI-powered credit scoring engine for Jisoti, a digital loan service that allows Selcom Pesa users to quickly access loans for emergencies, personal use, or small business activities. Selcom Pesa The solution determines credit limits for customers within Tanzania's leading digital financial platform.
Key Contributions:
- Built predictive analytics using alternative data sources including mobile money transaction history, app usage patterns, and repayment behavior
- Developed scoring algorithms that support loan amounts ranging from TZS 500 up to TZS 500,000 Selcom Pesa with flexible tenure options
- Created risk assessment models that analyze customers' recent three months mobile money transaction statements Selcom Pesa to determine loan eligibility increases
- Implemented real-time scoring APIs integrated with the Selcom Pesa mobile application
Impact: Enabled instant, paperless loan decisions for thousands of Tanzanian users, supporting financial inclusion across the country through Tanzania's largest cross-segment financial and payment services provider.
Developed a credit scoring engine powered by Machine Learning models to support stock financing for Dawa Mkononi, a leading Tanzanian B2B health-tech company revolutionizing the pharmaceutical supply chain across East Africa. The platform enables pharmacies, clinics, and health centers to order medicines, access embedded credit, and receive timely deliveries backed by real-time inventory data
The credit scoring solution enables health facilities to access financing for pharmaceutical stock procurement, reducing cash flow constraints and improving medicine availability across Tanzania. The company uses FinTech options to streamline payments, Empower Africa and support healthcare accessibility.
The Education Dashboard stands as a pivotal resource, offering users a comprehensive platform to observe and analyze examination results from various Tanzanian schools through insightful graphs. Focused on integrating CSEE results spanning from 2017 to 2022, this dashboard provides a valuable tool for understanding educational performance across the nation.
Note: Access link coming soon.
A speech recognition model in the swahili language using API to help farmers in Kenya and Tanzania (Arusha) ask questions orally related to livestock diseases via mobile app or web app.
Swahili tweets from Twitter were collected, annotated, and validated into three categories Positive, Negative and neutral.
The project delivered open, accessible, and high-quality text and speech datasets for low-resource East African languages from Uganda, Tanzania, and Kenya. The project focused on these languages: Luganda, Runyankore- Rukiga, Acholi, Swahili, and Lumasaaba. I led and supervise 3 communities in Tanzania to mainly focused on building Swahili datasets for both texts and speech.
Annotating and validating NER corpus for the Swahili language and other 19 African languages based on news data collected from news platforms.
Developed an ensemble of models that anticipate ten various forms of network intrusions, such as DoS, Shellcode, Backdoor, and others.
