HKUST Fintech Research Project
Data Analytics Research Assistant
Fine-tuned GPT/BERT models on NVIDIA DGX for domain-specific NLP, and led an HKMA industry report and a federated-ML project with SHAP explainability.
+10pp
Classification accuracy
NVIDIA DGX
Fine-tuning hardware
HKMA
Published industry report
What I worked on
- Fine-tuned GPT/BERT models on NVIDIA DGX for domain-specific text classification, lifting accuracy by 10 percentage points.
- Led an HKMA industry report using hierarchical clustering, from analysis through to publication.
- Led a federated machine-learning project for a public entity, using SHAP for model explainability and transparency.
NLPFine-tuningBERTFederated MLSHAP
