I'm a Master of Computing (Applied AI) student at Nanyang Technological University, Singapore. My background is in data science, spanning machine learning engineering and applied research on AI agents. What I care about most is getting AI out of notebooks and into real applications that are reproducible, explainable and safe to rely on.
Current focus
- Turning AI techniques into working products and research tools
- Agent systems that use tools, judge evidence and leave an auditable trail
- Rigorous ML evaluation: temporal validation, leakage control, interpretability
| Project | What it does | Stack |
|---|---|---|
| VizJudge | Judges which charts actually carry signal for a modelling task and turns that evidence into ML actions; works as a CLI, Python API and MCP server for host agents. | Python · MCP · scikit-learn |
| provenance-care-note | Provenance-first clinical care note prototype with role-based access, explainable priorities, audit history and PHI redaction, on synthetic data. | FastAPI · PostgreSQL RLS · WebSocket |
| uk-road-collision-severity-prediction | Out-of-time prediction of serious road collisions from 500k+ UK DfT records, comparing five classifiers with drift and calibration analysis. | LightGBM · CatBoost · geospatial |
| fractional-tde-classification | Tidal disruption event classification from LSST-style light curves using FFT-based fractional-order features. | XGBoost · LightGBM · SHAP |
- Languages: Python · SQL · JavaScript
- ML & data: scikit-learn · LightGBM · XGBoost · CatBoost · pandas · SHAP
- Agents & LLMs: MCP · tool-using agents · LLM-as-judge evaluation
- Engineering: FastAPI · PostgreSQL · pytest · GitHub Actions · uv