I'm Amrita Thavrani
I lead technical programs and write about AI in plain English: what the terms mean, how the systems work, and what it takes to ship them.
What was that again? makes AI simple for product and program managers who keep asking 'what was that again?' No bluffing required.
I've been in the room
The room where a model isn't behaving as expected and several teams have different ideas about what "done" means. A better prompt is sometimes the answer. Sometimes it's the wrong question.
I'm a Senior Technical Program Manager at Amazon, working on AI platforms. My lens is program leadership: connecting the technical work to clear goals, ownership, and delivery.
That's the perspective I bring here. Understanding a model matters. So does understanding the system around it and who's responsible when it stops working.
Why this blog exists
AI moves fast. The jargon moves faster.
I write the explanations I wish I'd had when I first encountered these ideas. What does the term mean? How is it different from the one that sounds almost the same? Where does it show up in a real system?
You'll find plain-English explainers, practical ways to think about AI programs, and the occasional small project built from scratch. The aim is simple: leave knowing enough to ask a better question, make a better decision, or explain it to someone else.
Less hype, more signal.
Start here
- Glossary - for the next time a meeting turns into alphabet soup
- Four Layers - find the layer that's broken before you rewrite the prompt again
- LLM Evaluation Is a Program Management Problem - making evaluation a working practice, with ownership beyond launch
Keep Engaging
Subscribe to get new posts by email.
Say hello: thavrani.amrita@gmail.com
These are my own views, not those of my employer.