Alexander Hepburn

AI Engineer | RAG · NLP · Conversational AI · Model Diagnostics

Data scientist and AI engineer specialising in NLP, information retrieval, and production machine learning. I work on problems end-to-end — from structuring the problem and running experiments to building and deploying the final system.

- Conversational AI Systems: Building sales and support agents that combine language model intent recognition with deterministic business logic. Systems are fully configurable — behaviour is adjusted through rules, not retraining. Integrations include Meta and WhatsApp.

- Retrieval-Augmented Generation: Building RAG pipelines where retrieval is the primary engineering focus — chunking strategy, embedding selection, semantic reranking — with strict generation constraints and source attribution throughout.

- Model Diagnostics: Diagnosing why models degrade in production. I run controlled ablation studies to isolate contributing factors — distribution shifts, data quality issues, label imbalance — and prioritise fixes by measured impact.

- Machine Learning & Data Science: Prototyping and deploying models for text and structured data, with experience in causal inference, A/B testing, and hypothesis-driven analysis.

Tech: Python · PyTorch · FastAPI · PostgreSQL · ChromaDB · AWS · PySpark