Architecture
Synapse vs Databricks: the decision is bigger than compute
A decision framework covering concurrency, operating model, governance, tuning overhead, cost behaviour and migration effort.
Coming nextDATA · CLOUD · AI · ENGINEERING IN THE REAL WORLD
Build systems.
Keep the lessons.
I’m Pravin. I write about data platforms, architecture decisions, migrations, automation, security, cost and the parts of engineering that become obvious only after something hits production.
DatabricksAzureMicrosoft FabricPlatform EngineeringFinOps
01Practical over theoretical
02Trade-offs over absolutes
03Reusable patterns over heroics
LATEST NOTES
Things worth writing down
The goal is not to publish every week. The goal is to publish when something is useful enough that I would want to find it again six months later.
Databricks
Building a metadata-driven Databricks workspace migration accelerator
Moving notebooks is easy. Migrating safely at scale is not. A practical design for inventory, dry runs, dependency checks, overwrite protection, auditability and rollback.
Read article →FinOps
Where Databricks cost quietly leaks
Idle compute, warehouse sizing and overlapping workloads — and what to measure before turning “cost optimization” into a slogan.
In the backlogBUILDING IN PUBLIC
Small tools for repetitive problems
I’d rather automate a boring engineering task once than explain the same manual runbook ten times.
01
databricks-toolbox
Utilities, experiments and reusable building blocks for Databricks platform engineering.
View on GitHub ↗02
syn-dbx-migration
Experiments and tooling around Synapse-to-Databricks migration patterns.
View on GitHub ↗03
The next useful thing
No filler project. This card gets replaced when a real problem earns a reusable tool.
Work in progressABOUT THE SITE
Not a résumé.
A working notebook.
Bits of Pravin is where I keep ideas that survive contact with real systems.
Most technical content shows the final answer. I’m more interested in what happened before the answer: constraints, bad assumptions, operational risk, failed approaches, and the trade-offs that shaped the design.
That makes this site deliberately narrow: data platforms, cloud engineering, automation, AI experiments and the occasional lesson that does not fit neatly into a diagram.
More about me and the site →BITS OF PRAVIN