AboutBuilding
Building
in public.
Java engineer — learning AI through real systems
Actively writing
Most writing about AI is either too shallow to be useful or too academic to be practical. This is the middle ground.
Who I am
I build backend systems in Java by day and pull apart AI models at night. SmartGNT is where I document what I actually learn — the parts that worked and the parts that didn’t.
Everything here is code-first. LLM internals — how attention, context windows, and fine-tuning actually work. Developer tooling — agents, RAG pipelines, and the things worth adding to your stack. AI in production — monitoring, evals, what breaks at 3am.
Now
Current focus
Writing a series on how LLM context windows actually work — why 128k tokens doesn’t mean what you think, and how to build systems around the limits. Building the reference implementation in Java alongside each post.
Stack
Java
Python
LangChain
OpenAI API
Spring Boot
Docker
PostgreSQL
Linux