For eighteen years I've been doing one thing: turning the messy, distributed, often-contradictory data inside large enterprises into reports leaders trust enough to make decisions on. Most of that work doesn't make for great Twitter content. It's semantic models, governance frameworks, security boundaries, performance tuning. The unglamorous mechanics that decide whether a Quarterly Business Review actually informs a quarter, or just fills slides.
Today I'm a Senior Architect, Data and Analytics at Tiger Analytics, leading engagements where the deliverable is not a dashboard but an analytics capability — the platform, the governance, the patterns, the team rituals. Recent work has spanned Microsoft Fabric implementations, Power BI tenant architecture, semantic model design at scale, and the messy organizational work of getting an analytics program adopted instead of merely deployed.
Before Tiger I spent twelve years at Infosys, working onsite in Toronto and Milwaukee and offshore from India. I delivered enterprise analytics on Power BI, MSBI, Azure, Snowflake, and Databricks — for some clients, the same semantic models I wrote in 2015 are still serving five hundred business users today. I'm prouder of that than of anything I shipped this year.
My background is unusual for this work. I started as a C/C++ engineer, building security and cryptography platforms and financial trading systems, before moving into BI in the SSIS/SSRS/SSAS era. The low-level engineering grounding is why I tend to think about query plans, storage engines, and Vertipaq compression earlier than most BI people do. It shows up in the work.
I write here at Analytics Foundry for the same reason most practitioners write: to think more clearly, to compress what I learn during real engagements into something durable, and occasionally to argue with the prevailing wisdom. The journal is opinionated. The case studies are honest about what didn't work. If anything here is useful to you, that's the entire point.
— Ankur