Data analyst and software engineer. I build data infrastructure, operational reporting systems, and NLP tooling, with a focus on problems that matter in underserved contexts.
I'm a software engineer and data analyst with 7+ years of experience building data systems, reporting pipelines, and operational tools, mostly in Somalia, for contexts where good infrastructure is scarce and getting it right actually matters.
My day job is at SOSTEC Technologies, where I run the software department and maintain a multi-module accounting and reporting platform serving 200+ customers. That means data validation, reconciliation workflows, database integrity across PostgreSQL, MySQL, and Oracle, and translating all of it into something non-technical stakeholders can act on.
Outside of that, I've been building tools that fill gaps I notice in the Somali digital ecosystem: a structured NLP corpus for Somali (a low-resource language with almost no digital linguistic data), and data assurance pipelines modelled on humanitarian delivery operations.
I hold a BSc in Statistics and Planning, the Google Advanced Data Analytics certificate, and a Udacity Data Analysis Nanodegree. I'm currently pushing into ML engineering; the progression from analytics to modelling is the next logical step, and the Somali NLP work is where that gets interesting.
End-to-end digitization of the Qaamuuska Af-Soomaaliga, the most comprehensive monolingual Somali dictionary in existence (~900 pages, Roma Tre University Press, 2012). Transforms an inaccessible academic PDF into a queryable, web-accessible dictionary and NLP-ready corpus. One of the first structured digital datasets for the Somali language, directly enabling downstream work on tokenizers, POS taggers, lemmatizers, and LLM fine-tuning.
A beneficiary data assurance pipeline for humanitarian delivery operations in Somalia. Ingests distribution files, runs schema validation, deduplication screening, chronology checks, and status audits, and then produces a reconciliation report and anomaly register for Area Office review. Built to mirror real operational concerns: targeting integrity, delivery cycle tracking, and caseload reconciliation. Includes a full CI/CD test suite via GitHub Actions.
An interactive operational dashboard for the World Food Programme (WFP) that tracks beneficiary aid delivery cycles, calculates real-time completion rates per region, and detects duplicate payout conflicts. Features an interactive date simulator to flag logistics delays, horizontal bar charts with custom tooltips via Chart.js, a sorting/paginated record explorer, and custom CSV dataset importing.
When assistance reaches the wrong person twice, or fails to reach someone at all, the consequences are not line items on an audit report. A look at what deduplication actually means in the field, and why the system's job is to generate verified queues, not make decisions.
More articles coming. Topics: Somali NLP, data systems in low-resource environments, ML for humanitarian applications.
I'm open to conversations about data systems, humanitarian data work, Somali NLP, and technical roles where the work has genuine stakes. Drop me a line if any of that overlaps with what you're building.
abdallammud@gmail.com