I’m Ram.

I’ve spent twelve-plus years in data and software engineering.

Most of it is enterprise data work: years of ETL and warehousing (the on-prem, Informatica-era kind), then cloud data platforms — Azure Synapse, Data Lake, Databricks — and lately architecting on Microsoft Fabric and leading a migration off SSIS. A lot of the job is taking messy business logic and turning it into something that can be trusted: pipelines that can rerun, models that can be explained, and systems that do not collapse the moment a new column, source file, or exception shows up.

The less obvious half is where a lot of the writing comes from. My first master’s was in computer science — game engines, graphics, CUDA — which is why the oldest notes here are about GPU ports and netcode; a later master’s in data science pushed me toward the data career. That mix, plus a stubborn homelab habit, is the real lens: enterprise data architecture, hands-on AI and GPU work, and hardware I can actually hold, all feeding each other.

I’m not interested in pretending every new tool is a revolution. Most are useful in some places, painful in others, and dangerous when people stop asking what they’re actually doing. That’s usually the lens I write from.

The informal side

Outside formal work, I tend to experiment with systems.

Sometimes that means local AI models and homelab infrastructure. Sometimes it is 3D printing, Raspberry Pi, OpenWrt, pfSense, Proxmox, Unraid, OpenMediaVault, embedded boards, LTE/5G modules, or random IoT devices. Sometimes it is photography, small utilities, or some weird workflow problem that refuses to leave my head.

I do not treat all of these as serious credentials. They are more like practical experiments that shape how I think.

When you spend enough time debugging routers, SBCs, storage boxes, sensors, data pipelines, and flaky software, you start developing a low tolerance for systems that only work in theory.

What I care about

A handful of beliefs run under everything here: systems should stay understandable, AI should assist while deterministic code makes the decisions, small tools should resist becoming platforms, and software should respect the user’s ownership of their own files and data.

That is the short version. The long version — the reasoning and the scars behind each one — lives on the philosophy page.

What you will find here

This site is a mix of data engineering field notes, Microsoft Fabric experiments, AI-native tool design, local AI and homelab notes, product architecture thinking, solo-builder ideas, knowledge system experiments, hardware and infrastructure tinkering, and occasional domain deep dives.

Some of it will be polished. Some of it will be closer to a working notebook.

That is intentional. I would rather publish useful thinking while it is still alive than wait until every idea is perfectly packaged.

If you want to get in touch, see contact.

Everything here is personal opinion and independent writing. It does not represent any employer.