About Me

I am a final-year PhD candidate at the Artificial Intelligence Applications Institute (AIAI), School of Informatics, The University of Edinburgh, advised by Prof. Tiejun Ma. I previously completed an MRes at the Institute for Language, Cognition and Computation (ILCC), The University of Edinburgh, advised by Prof. Shay Cohen. I am also currently an Intern of Technical Staff at Cohere.

My research has two closely connected strands. The first develops post-training and adaptation methods for reliable LLM agents. I study how language models can learn from feedback, improve through self-play, use memory, retrieve relevant evidence, and act robustly in long-horizon settings. Representative work includes controllable LLM behaviour through SEKA, temporal and agentic memory through RoMem, and robust retrieval/ranking through TSPRank.

The second strand applies these methods to FinAI, where reliability, distribution shift and long-horizon evaluation are central. My work studies LLM-based investment agents, trader risk ranking, synthetic transaction generation, financial backtesting and look-ahead-bias mitigation. This includes financial AI systems such as FINSABER, FinCAD, PA-RiskRanker, and SynthRank.

Across these two strands, my ultimate goal is to make algorithmic advances that matter, and deploy them reliably in high-stakes real-world decision-making.

I have multiple first-author papers at leading venues, including ICLR, KDD, EMNLP, TOIS, AAAI and EACL. I also collaborate on broader agentic AI research, including world modelling and vision-language-action systems, with researchers across academia and industry. I also have interned at Tencent Lightspeed Studio UK, Huawei Technologies Research & Development (UK) Ltd, 2012 Laboratories, and The Alan Turing Institute, and have contributed to teaching machine learning and natural language processing courses at the The University of Edinburgh.

I will be on the 2027 job market and am currently looking for full-time Research Scientist or Applied Scientist roles starting in 2027, as well as research internship opportunities before then.