Why RAG gives wrong answers (and how to fix retrieval failures)
Two common RAG failure modes diagnosed and fixed with real code, plus a toolkit for chunking and retrieval strategies and when to reach for each.
Series · Learning AI Out Loud
How to make AI answer questions about your documents, building RAG from scratch
Build a working RAG pipeline in about 60 lines of Python. Chunk documents, embed them, search by meaning, and generate grounded answers with Amazon Bedrock.
Why does AI forget what you said, and how to fix it
Why your AI gets worse the longer a conversation runs, and the strategies that keep it on track.
Bigger AI models aren't always better. Here's how to actually choose.
A recipe demo plus the cost-to-scale trade-offs for picking the right model for the job.
Why does AI lie? Hallucinations explained simply
What hallucinations actually are, why models make things up, and simple demos that show it happening.
What even is AI? I took a break, had to relearn everything
Coming back from a break and feeling behind. Building AI mental models from first principles, no jargon.
More Writing
The AWS Cleanup We Keep Putting Off (I Let an Agent Do It)
The tidy-up nobody wants to do. I handed it to an agent and watched it work through my AWS account.
Theatre Kid to Developer Advocate at AWS
The career story. From a one-way flight and a leap of faith to developer advocacy at AWS.
Also cross post on dev.to - https://dev.to/rohini_gaonkar
How to stop overspending - AWS Cost Optimisation tips
Sharing with you typical cost savings tips I have shared with customers as a Solutions Architect. These are some of the obvious (but often ignored) items that customers can do to save cost as they build and grow on Amazon Web Services (AWS).