Head of AI Research · PhD in Computer Science (University of Manchester)
I build production LLM systems: RAG, agents, and the evaluation and guardrails that make them safe to ship. Over 9+ years I've taken ML from research to production in edtech, healthcare and industrial IoT. Today I lead applied AI for a global learning platform used by millions of students.
- 🤖 LLM applications & agents: RAG, semantic search, multi-agent orchestration (LangGraph, AWS Bedrock)
- 🧪 Evaluation & responsible AI: eval pipelines, guardrails, response validation, safe deployment
- 🎯 Adaptive learning: reinforcement-learning recommendations, uncertainty-aware ability/difficulty rating
- ☁️ ML platforms: end-to-end MLOps on AWS (ECS, Lambda, SQS, CDK), Kubernetes, Kubeflow
- Led architecture of an AI Teacher Assistant that lets teachers analyse class and student performance in natural language
- Shipped production RAG and agent systems for learning-gap analysis, lesson planning and content recommendations, with evaluation and guardrails built in
- Built privacy-preserving NLP / GenAI integrated with EHR platforms in regulated healthcare
Most of my production work is in private repositories. The projects below are open-source and research work.
- Towards Knowledge-aware Few-shot Learning with Ontology-based n-ball Concept Embeddings, IEEE ICMLA 2021 · code
- Visual-semantic Embedding Models Informed by Structured Knowledge, STAIRS @ ECAI 2020
Python PyTorch LangChain LangGraph Langfuse FastAPI AWS Bedrock ECS/Fargate Lambda CDK Docker Kubernetes Kafka OpenSearch Pinecone
