Ten languages, one habit: wherever I work, the model goes in the product, not in a notebook.
I'm an AI developer, building custom MCP servers that connect models to real systems and RAG pipelines over private data. The AI part isn't new. Back when I shipped Android apps and fullstack web products, I was already putting ML inside them (on-device segmentation in Kotlin apps, model-backed features in websites), years before "agent" was a job requirement. What changed is the delivery mechanism, not the work.
The repos below are the receipts. Have a look around.
I teach what I build: AI and ML, MCP servers and RAG, web development, design, and content creation. If you learn better by watching someone actually make the thing, that's the format. Find it on Instagram.
I write about agent skills, Android, and ML on Medium. The posts are build logs, not thought leadership.
If something here is useful to you, a star helps other people find it.