Terry Migwi

I’m an AI Engineer and Data Scientist with 5 years of experience spanning data analysis, machine learning, and applied AI. I start with predictive models and analytics pipelines, then evolve them into production-grade systems—especially RAG applications, LLM-powered agents, and workflow automation tools.

In my most recent role, I protected approximately £450,000 in annual revenue by building a risk scoring model that identified at-risk learners and enabled targeted interventions, supporting a 90% retention rate across 60 apprentices. I also saved significant time by deploying a Gemini-powered extraction pipeline to auto-populate structured learner data, and by building a RAG-like FAQ system that handled around 50 apprentice queries per day with grounded responses.

I’m motivated by end-to-end ownership: from modeling and feature selection to building reliable services with FastAPI, async task processing, and resilient retrieval pipelines. I’ve also delivered segmentation models, semantic search over document corpora, and semantic credit-scoring improvements—always focused on measurable outcomes, faster iteration, and systems that hold up under real usage.