I help enterprises turn GenAI ambition into infrastructure that actually ships — and I show my work.
About vExpose
I’m Deepak Waghmare, and this is where I think out loud about building enterprise technology that survives contact with production.
My one conviction — the thread through everything here: AI fails in the operating model, not the model. The tool is not the transformation. A better model doesn’t fix a broken system; it just takes the wrong action more fluently. So the work that matters isn’t chasing the newest model — it’s the architecture, the automation, and the operating discipline that decide whether any of it actually holds up.
For more than two decades I’ve worked deep in the layers most people never see: disaggregated and GPU-dense platforms, storage and performance engineering, virtualization, and the DevOps and infrastructure-as-code that hold it all together — including work across the Dell Technologies ecosystem. Nothing here is theory I read about; it’s what I’ve built, benchmarked, and occasionally broken.
What “Learnings From Tech-Exposure” means
vExpose is my public workbench. The premise is simple: most technical writing shows you the strategy but hides the wiring. Here I try to do both — state the call at the altitude a technology leader has to make it, then prove it in code, architecture diagrams, and real benchmark data below. If a post doesn’t have something you could actually build from, it isn’t finished.
You’ll find three things here:
- GenAI × enterprise infrastructure — reference architectures for putting AI into real systems: agentic workflows, RAG and enterprise search, the platforms underneath GenAI.
- Platform & architecture strategy — the build-vs-buy calls, the tradeoffs, and the operating-DNA decisions that determine whether a platform scales or stalls.
- Automation & IaC in practice — Ansible, Kubernetes, Python, ServiceNow, REST APIs, and IaC security, with working code you can lift.
Running underneath all of it is a habit I’d rather over-index on than under: measure it. Performance and benchmarking — the fio/vdbench, real-numbers kind — is the discipline I trust most, and it shows up across everything I write.
How I think about leading under uncertainty
Technology leadership isn’t only architecture; it’s judgment when the picture is unresolved. I try to practice conviction with a review date — a direction stated strongly enough to act on, dated explicitly enough to revise when the evidence changes — instead of the safe fog of professional neutrality, where you say a great deal and commit to nothing. When the data is ambiguous, the job is interpretive courage: looking at the picture everyone can see and saying what you think it means. That’s why this blog has opinions, and why they come with my name on them.
Let’s connect
If you’re building GenAI into enterprise systems, wrestling with a platform decision, or just want to compare notes on something here, I’d like to hear from you.
LinkedIn · X / @waghmaredb · GitHub
Subscribe to get new reference architectures, build logs, and benchmarks in your inbox — no noise, just the work.