Research Associate and ML Engineer at the University of Surrey. I build production data systems, LLM-powered pipelines, and full-stack software, and publish research across machine learning, energy forecasting, and transport.
PhD from Heriot-Watt in deep learning for residential energy demand forecasting. Before that: geology at the University of Tehran, with applied work on tunnelling and groundwater.
KTP Associate, Surrey Business School / Cycle Confident (Feb 2024 – Present)
Building intelligent analytics infrastructure for active travel.
- Full-stack analytics platform (React, Django, PostgreSQL) serving 50+ internal users and 200+ external clients
- Deployed a fine-tuned 7B LLM on 2M+ domain-specific documents for natural language BI querying
- Agentic AI system (Orbit) with RAG-based SQL generation, LLM evaluation framework, and access-controlled data scoping
- ETL pipelines processing 500K+ daily records on Docker and Railway
Mobile Developer, Radzi (Oct 2025 – Present)
Active travel tracking app (React Native, Expo) for cyclists and pedestrians, feeding usage data to local councils.
Co-founder, Zynthia
Research platform that keeps systematic literature reviews continuously updated.
Co-founder, Veticity
Platform for improving academic manuscript submissions before journal review.
169 citations · h-index 5 · i10-index 4
Published in Expert Systems with Applications, Energy and Buildings, Sensors, Journal of Computer Information Systems, and Bulletin of Engineering Geology & the Environment.
AI & LLMs: Foundation model fine-tuning · RAG · LangChain · LangGraph · LlamaIndex · Ollama
ML: Transformers · CNNs · Time Series · MLOps · NLP
Full-Stack: React · React Native · Next.js · Django · REST APIs
Languages: Python · JavaScript · SQL · R · LaTeX
Infrastructure: PostgreSQL · Docker · Google Cloud · Railway · Firebase
Visualisation: Power BI · Plotly · D3.js