I build some AI,
extract business insights,
and create something crazy.
Welcome to my personal learning lab & project portfolio. I like solving real problems, exploring messy datasets, building local AI tools, and documenting what I learn along the way.
CONNECT WITH ME
Systems & Data Engineering
cheeserag
Privacy-first, fully offline local RAG workspace & NotebookLM alternative powered by embedded C++ vector DB.
pomaidb
Predictable, embedded multimodal vector database & offline RAG engine for Edge AI in C++20.
Customer Segmentation
Interactive Next.js analytics platform & K-Means clustering model (k=8) analyzing 15k customer records with insight matrices.
Fixago
Book fast & reliable home repair services in Vietnam with 24/7 AI-powered instant booking.
Latest from the blog
Why Does Segmentation Need Multiple Scales?
I started looking at segmentation architectures from a different angle: the real problem is not simply predicting a class for every pixel, but preserving enough spatial information while building representations with a sufficiently large receptive field.
A Larger Context Window Does Not Remove the Need for RAG
A large context window sounds like it should make retrieval unnecessary. But after working with constrained local models and explicit retrieval pipelines, I started looking at the problem differently: context capacity and context quality are two different engineering problems.
Engineer. Analyst. Product Builder.
I'm a multidisciplinary engineer and analyst passionate about building scalable software systems and turning complex data into clear business decisions. My work spans AI systems, backend architecture, statistical analytics, and product strategy.