Hi, I'm Ammar.
I'm a machine learning engineer in Kuala Lumpur, Malaysia, with a focus on MLOps: getting models out of notebooks and into services that people can rely on.
That means standard pipelines, monitoring, and deployment that scales. Lately my attention is on AI platforms and agents: how they call tools safely, who they act for, and how you can tell when they quietly go wrong. I write about it here, and I build small open-source tools and live demos to test the ideas.
- Based in
- Kuala Lumpur, Malaysia
- What I do
- Machine learning engineer, focused on MLOps and model deployment
- Working on
- AI platforms and agents: tool calls, identity, observability and reliability
- Building since
- 2017, in public on GitHub
Focus
- MLOps
- Model deployment
- AI platforms
- AI agents
- Computer vision
- Retrieval-augmented generation
A few things I've built
- agent-gateway-playground Break an AI agent on purpose, and watch every step.
- mlops-toolbox A model that carries its own contract from training to production.
- pos-classifier An end-to-end text classification service with a human-review loop.
- RAG-Toolbox Chat with a PDF using retrieval-augmented generation.
I've been building in public on GitHub since 2017: computer vision and NLP experiments, deployment pipelines on Azure and GitHub Actions, and MLOps and RAG toolkits. All projects ›