Ahmed Younes
Navigating data into decisions
I work with organisations to navigate their data challenges — from raw unstructured text to intelligence that drives real decisions. I work at the intersection of data science, natural language processing and engineering — combining research depth with hands-on building and strategic consultancy. I've worked with organisations including the Gates Foundation, FCDO, ISD, BBC Monitoring and Ofcom, and I guide professionals across 50+ organisations in applying AI and data science in practice.
Available for consulting, research & building AI solutions
Selected Work
Projects
Emotech · Contract
LLM-as-a-Judge Evaluation Framework for Conversational AI
Built an automated evaluation pipeline for a customer service voice AI — GPT generates adversarial test scenarios, drives live conversations, and judges outputs across design bugs, logic bugs, and quality metrics. Scaled from 7 to 42 test cases across 14 categories.
View Details →AxessAll · Self-employed
AI Strategy & Roadmap for a National Accessibility Platform
Led AI strategy and roadmap planning for a national government digital accessibility programme — a seven-layer platform architecture, a three-agent automation model, and direct contribution to bid and proposal development.
View Details →EMIF / ISD
Pro-Kremlin Influence Network Mapping
Mapped pro-Kremlin influence networks across 7.8 million Telegram posts in France, Germany and Italy.
View Details →Gates Foundation
Global Fund Advocacy Evaluation
Evaluated social media advocacy for the Global Fund's 7th replenishment across 5 markets and 43,000 posts.
View Details →Work History
Experience
AxessAll
QA Ltd
Emotech
CASM Technology
Industries & Sectors
Industries & Sectors
Through my role with QA, I've engaged with teams across 50+ organisations spanning multiple sectors — helping them incorporate data science and AI into their work.
Doctoral Output
Research
PhD Doctoral Output — University of Sussex, 2025
DeformAr
A diagnostic framework for NER evaluation across Arabic and English — introducing an entity-level error taxonomy and cross-lingual benchmarking protocols to surface where and why NER models fail on morphologically complex languages. DeformAr moves beyond aggregate F1 scores to provide actionable diagnostics for low-resource Arabic NER.
Analytical Toolkit
Methods
Topic Modelling Recipes
A practitioner's guide to topic modelling — checklists, investigations, and methodological decisions accumulated across multiple production projects.
Multilayered Topic Modelling
Hierarchical topic discovery from broad themes to granular sub-topics, enabling scalable narrative analysis across large corpora.
NER Evaluation Workflow
Three-stage benchmarking framework across low and high resource languages, surfacing systematic error patterns at entity level.
Machine Translation Evaluation
Comparative metric framework using BERTScore as primary signal, combining automatic and human evaluation for robust translation quality assessment.
Get in Touch