and derived risk indicators

• Dashboard-ready SOC, temperature and fault visualisations

"},{"id":"589kd7l","name":"Aged Care Invoice Data Quality","meta":"Business Process Analytics · 2026","role":"Data & Business Process Analyst","tags":["Data Validation","Reconciliation","Excel","Process Mapping","Aged Care"],"body":"

Developed a structured invoice-validation and funding-reconciliation workflow for aged care services.

Key Results

• Reconciled 660+ invoice records

• Identified and escalated 11 exceptions

• Mapped 50+ providers and stakeholder relationships

• Improved traceability across payment and funding workflows

Anonymised professional case study — no client or invoice data shown.

"}],"tagline":"From raw data to reliable, decision-ready systems.","awards":["NetEase Business Data Analyst — High Distinction"],"footer":"Python · SQL · AWS · S3 · Glue · Athena · Data Quality"},{"id":"ito9q32","title":"Applied AI & Machine Learning","items":[{"id":"dbn544u","name":"SDG Knowledge System","meta":"LLM & RAG · 2025","role":"Data Scientist · Database & Retrieval","link":"https://sdg.unswzoo.com/","tags":["Python","Django","LLM","RAG","FAISS","Milvus","LoRA"],"body":"

Co-developed an AI-powered knowledge system for exploring the United Nations Sustainable Development Goals. Led database and ORM design, built core data APIs, and contributed to semantic retrieval and domain-specific LLM evaluation.

Key Results

• 22% improvement in retrieval recall

• 30% improvement in data-workflow efficiency

• Estimated 18% improvement in response quality

"},{"id":"4e6bmoe","name":"Faster R-CNN Pest Detection","meta":"Computer Vision · 2025","role":"Data Scientist · Model Evaluation","tags":["PyTorch","Faster R-CNN","ResNet-50 FPN","Albumentations"],"body":"

Developed and evaluated an agricultural pest-detection model across 13,140 images. Applied data augmentation and corrected the original AUC evaluation logic.

Key Results

• mAP50: 0.760 → 0.801

• AUC: 0.823 → 0.873

• Recall: 0.700 → 0.849

"}],"tagline":"Applying AI to search, vision, and predictive decision-making.","awards":["UNSW Master of Information Technology — Distinction"],"footer":"PyTorch · scikit-learn · RAG · FAISS · Milvus · XGBoost"},{"id":"vrjcxvk","title":"FinTech & Business Intelligence","items":[{"id":"gwwf7zt","name":"Privacy-Preserving AML & Anti-Fraud","meta":"FinTech Risk Analytics · 2020–2023","role":"Data Engineer & Innovation Lead","tags":["AML","Federated Learning","Privacy Computing","SQL","Data Modelling"],"body":"

Built privacy-preserving financial-crime models in collaboration with regulatory and internal project teams.

Key Results

• Nine AML data models

• 85 federated anti-fraud feature models

• 40% less manual data-preparation work

• Estimated 18% reduction in investigation workload

Anonymised professional case study — no confidential data shown.

"},{"id":"reo8t29","name":"Regulatory Reporting Automation","meta":"Enterprise Data Engineering · 2020–2022","role":"Data Engineer","tags":["SQL","ETL","Reconciliation","Automation","Regulatory Reporting"],"body":"

Transformed fragmented banking data and regulatory rules into repeatable reporting workflows.

Key Results

• Integrated 68 fields across 45 tables

• Automated 15 central-bank reports

• Improved reporting efficiency by 60%

• Reduced turnaround time by 70%

Anonymised professional case study — no confidential data shown.

"}],"tagline":"Trusted data products for risk, reporting, and commercial growth.","awards":["2021 Outstanding Product Promotion Innovation Award","2022 BOC Micro-Lesson Competition — Third Prize"],"footer":"SQL · AML · Privacy Computing · Federated Learning · SmartBI"},{"id":"922yzru","title":"Creative Technology","items":[{"id":"op2yetj","name":"Interactive 3D Personal Portfolio","meta":"Creative Web Development · 2026","role