LinkedIn · Explore my repositories
I'm a technology professional based in Singapore, interested in how enterprise systems, thoughtful architecture, and open-source tools can turn AI ideas into useful applications.
This GitHub is my workshop: a place to build workbenches, explore design choices, and learn through hands-on experiments in AI, data engineering, and cloud systems.
The questions that interest me: How does it fit into the wider system? How can we observe it? And how do we know it works?
| Area | What I like exploring |
|---|---|
| Enterprise systems | Connecting applications, data, and operational workflows |
| Architecture | Clear boundaries, integration patterns, and practical trade-offs |
| Open source | Building with tools that can be inspected, adapted, and run locally |
| AI | RAG, agents, model adaptation, evaluation, and assurance |
A local lab for exploring how AI applications can be tested and reviewed, bringing together threat modelling, evidence review, and scorecard-based reporting.
Explore: AI assurance · security evaluation · reproducible experiments
Tools: ASQI Engineer, Garak, Ollama, FastAPI, Python
A property operations lab exploring how retrieval and tools can connect AI applications with documents and structured data, with observability in the workflow.
Explore: RAG architecture · data integration · tracing
Tools: OpenRAG, S3, Google Drive, SQLite, OpenTelemetry, LangSmith
An experiment with FLAN-T5 dialogue summarization, comparing full fine-tuning with parameter-efficient fine-tuning and evaluating results using ROUGE.
Explore: model adaptation · evaluation · fine-tuning trade-offs
A notebook exploring collaboration between AI agents using CrewAI, including research, content generation, and task orchestration.
Explore: agent collaboration · workflow automation · orchestration
More from the learning notebook
- Safer dialogue summarization with PPO and PEFT — exploring reward-based approaches to reducing toxicity in generated summaries.
- LangChain application development — course exercises and experiments with LLM application building blocks.
- How can RAG systems connect to enterprise data while remaining understandable and observable?
- What evidence helps us assess an AI application's reliability and security?
- Where do agents add value to a workflow, and where is a simpler approach enough?
- How can local and open-source tools make experiments easier to reproduce?
Interested in enterprise architecture, open-source systems, or practical AI? I'd enjoy exchanging ideas.