Shafayet Ahmad Kanon
PhD researcher, Centre for Safety Resilience and Protective Security (CSRPS), School of Computing and Mathematical Sciences, University of Greenwich, London.
I am a computer science PhD researcher at the University of Greenwich working on reliable and accountable artificial intelligence. My research spans large language model evaluation, retrieval-augmented generation, computer vision, knowledge representation and human oversight of AI systems. I am particularly interested in building AI systems whose outputs can be evaluated, traced and used responsibly in high-stakes professional workflows.
My PhD uses fire safety engineering as a safety-critical application domain. I study how AI and software systems can support technical-document analysis, regulatory reasoning, drawing and geometry interpretation, simulation workflows and engineering decision-making without obscuring uncertainty or professional accountability. I am supervised by Dr Peter Lawrence and Dr Asim Siddiqui and funded by the M³4Impact / E3 doctoral studentship.
Before beginning the PhD, I completed an MSc in Computer Science with Distinction at Greenwich. My dissertation developed a real-time computer-vision system for detecting fire and smoke from video, incorporating human-in-the-loop alarm verification. I also hold a BTech in Computer Science and Engineering from KIIT University. My software engineering experience includes TypeScript, Node.js, REST APIs, containerisation, CI/CD and the development of research software prototypes. I am open to collaboration in AI evaluation, RAG, computer vision and safety-critical software systems.
Research interests
- Reliable and evaluable AI: faithfulness, hallucination and unsupported-claim measurement, citation correctness, calibration and human–model agreement
- Large language models and retrieval-augmented generation over technical, regulatory and scientific documents
- Data-centric machine learning: dataset construction, corpus quality, deduplication, contamination and evaluation design
- Computer vision and multimodal AI for video, documents, diagrams and architectural or engineering drawings
- Knowledge representation using graphs, ontologies, RDF and Neo4j for reasoning and information integration
- Human-in-the-loop, human-on-the-loop and agentic AI systems in high-stakes environments
- Research software engineering, reproducible experimentation and auditable AI-assisted workflows
- Applications of AI to openBIM/IFC, regulatory compliance and performance-based fire safety engineering
News
- — Gave a turbo talk on AI in fire engineering workflows at the NRIPS Symposium 2026, University of Greenwich.
- — Commenced the PhD at the University of Greenwich, funded by the M³4Impact / E3 doctoral studentship.
Selected writing and outputs
- Kanon, S., Lawrence, P., Siddiqui, A. “A Risk-Informed Classification Framework for Artificial Intelligence Autonomy in Fire Safety Engineering Workflows.” Manuscript in preparation with supervisors.
- — “AI in Fire Engineering Workflows: Opportunities, Challenges and Trustworthy Integration.” NRIPS Symposium 2026, University of Greenwich. Internal university research symposium.
The full list is on the publications page.
Contact
The best way to reach me is email, at shafayet.ahmad1@gmail.com. I am also on GitHub, LinkedIn and ORCID.
I also write informally on Medium and publish videos onYouTube. Neither is research output.